Page 876«..1020..875876877878..890900..»

FBI Identifies Lazarus Group Cyber Actors as Responsible for Theft … – Federal Bureau of Investigation

The FBI is issuing this release to warn the public regarding the theft of approximately $41 million in virtual currency from Stake.com, an online casino and betting platform. The FBI has confirmed that this theft took place on or about September 4, 2023, and attributes it to the Lazarus Group (also known as APT38) which is comprised of DPRK cyber actors.The FBI investigation has revealed that DPRK cyber actors moved stolen funds associated with the Ethereum, Binance Smart Chain (BSC), and Polygon networks from Stake.com into the following virtual currency addresses:

These sameDPRK actors are also responsible for several other high-profile international virtual currency heists. In 2023 alone, DPRK cyber actors have stolen more than $200 million. This amount includes, but is not limited to, approximately $60 million of virtual currency from Alphapo and CoinsPaid on or about July 22, 2023, and approximately $100 million of virtual currency from Atomic Wallet onor about June 2, 2023.

The FBI previously provided information to the public regarding the DPRKs attacks againstHarmonys Horizon bridgeandSky Mavis Ronin Bridge and put out acybersecurity advisory on TraderTraitor.In addition, the U.S. Department of Treasurys Office of Foreign Assets Control(OFAC) sanctionedthe Lazarus Group in 2019.

Private sector entities are encouraged to review the previously released Cyber Security Advisory on TraderTraitor and examine the blockchain data associated with the above-referenced virtual currency addresses and be vigilant in guarding against transactions directly with, or derived from, those addresses.

The FBI will continue to expose and combat the DPRKs use of illicit activities to generate revenue for the regime, including cybercrime and virtual currency theft. If you have any information to provide, please contactyour local FBI field officeor the FBIs Internet Crime Complaint Center atic3.gov.

The rest is here:
FBI Identifies Lazarus Group Cyber Actors as Responsible for Theft ... - Federal Bureau of Investigation

Read More..

Insurance industry grapples with rising cyber crime threat, climate … – Reinsurance News

In the Insurance Banana Skins 2023 report, a joint effort by PwC and CSFI, the global insurance industrys primary concerns have been revealed, with mounting fears over cyber crime and escalating worries regarding climate change.

The insurance industry faces its foremost challenge in the form of cyber crime. This threat encompasses the potential theft of sensitive data, phishing, and ransomware attacks.

Notably, cyber crime has held its top position as the most significant global risk since the 2021 report.

Insurers are navigating an increasingly intricate digital landscape characterised by new technologies, cloud computing, and extended supply chains. This complexity amplifies their vulnerability to cyber threats.

To counter this, the report underscores the urgent need for proactive cyber security measures integrated into both business and IT architecture.

Insurers voice concerns regarding the misuse of artificial intelligence as a potent weapon for breaching security, with a specific focus on state-sponsored cyber crime.

Rising costs associated with mitigating cyber threats are causing unease among insurers. The expansion of the insurance industrys IT ecosystem, driven by the proliferation of internet-connected devices and reliance on cloud and third-party services, heightens the challenge of defending against cyber attacks.

Climate change has ascended to a prominent position, ranking among the top three concerns for insurers. This reflects the growing immediacy and impact of climate-related risks.

Governments and regulators are increasingly urging insurers to enhance risk quantification and management in alignment with broader Environmental, Social, and Governance (ESG) agendas.

Regulation remains the primary concern for life insurers, closely followed by cyber crime. The sector grapples with additional challenges related to interest rate risk, macroeconomic factors, and investment performance.

While technological change is a shared concern, the risk from artificial intelligence ranks lower, and climate change is of lesser concern compared to non-life and reinsurance sectors.

Property and casualty insurers in the non-life sector are primarily concerned about climate change, followed by operational risks stemming from cyber crime. High-level concerns include technology, artificial intelligence, and regulation.

Change management challenges take higher precedence in this sector compared to the life insurance sector, although macroeconomic risk is of lesser concern.

Following the record-breaking global heatwave in July 2023, climate change has risen as a significant risk for reinsurers. They share concerns about cyber crime, technology, and artificial intelligence with other sectors.

Recruitment issues are common, and reinsurers are unique in placing de-globalisation among their top ten risks, reflecting the global nature of their business and concerns about protectionism.

Continue reading here:
Insurance industry grapples with rising cyber crime threat, climate ... - Reinsurance News

Read More..

Iranians hit by internet curbs in year since protests – FRANCE 24 English

Amini, a 22-year-old Iranian Kurd, died on September 16, 2022 after her arrest in Tehran for an alleged breach of the Islamic republic's strict dress code for women.

Her death triggered months of nationwide protests in which hundreds of people were killed, including dozens of security personnel, before authorities moved to quell what they branded "riots", including restrictions on the use of social media.

Alma Samimi, who makes a living by selling leather bags online, said her business had suffered since the blackout came into force.

"The damage has been irreversible," she said, with profits linked to her Instagram account -- which had amassed thousands of followers -- nosediving 80 percent.

"Online interactions have dramatically dropped since last year," she said.

The restrictions, which affected online activity including on Instagram and WhatsApp, come as millions of Iranians struggle to make ends meet, grappling with an economic crisis marked by soaring inflation and the Iranian rial's sharp decline.

Iran's economic woes were significantly compounded by Washington's 2018 decision to reimpose sanctions on Tehran after then-president Donald Trump unilaterally withdrew the United States from a landmark nuclear deal.

"We cannot plan for the future anymore," Samimi lamented.

Some 1,200 business owners have since called on Iran's President Ebrahim Raisi to lift the restrictions, according to local media.

To keep her business afloat, Samimi resorted to using virtual private networks and anti-filter tools to bypass the censorship.

Daily demand on VPN services in Iran rocketed to "3,082 percent higher than before the protests" last year, according to Top10VPN, a British-based digital security advocacy group.

Mohammad Rahim Pouya, a psychologist who offers online therapy sessions, said he lost more than 50 percent of his clients in the early days of the blackout.

The 32-year-old said he could maintain communication lines with clients in Iran but it was more challenging with those abroad.

For him, the use of anti-censorship tools was "an added cost and probably has security flaws".

"But what are the options?"

Finding ways around internet curbs has become commonplace in Iran where authorities have repeatedly blocked access over the years at times of unrest.

In 2009, access to social media networks was blocked during mass protests -- which became known as the Green Movement -- following a contentious presidential election that saw the populist Mahmoud Ahmadinejad take office.

Popular social media networks including Facebook and X, formerly known as Twitter, have since been blocked.

A decade later, harsher curbs were imposed after protesters took to the streets over a government decision to raise fuel prices by 200 percent.

In March, Iran's telecommunications minister Issa Zarepour urged foreign companies to introduce representative offices in Iran, saying "no one wants to limit the internet and we can have international platforms".

Meta, the American giant that owns Facebook, Instagram and WhatsApp, has said it has no intention of setting up offices in the Islamic republic, which remains under crippling US sanctions.

The internet curbs have been costly.

The government spent $773 million in 2022 alone to impose them, making it the second biggest spender on restrictions after Russia, according to Statista, a Germany-based data collection website.

In February, Iran's reformist Shargh daily reported that the country's internet service providers suffered 40 percent losses because of the restrictions.

With the bans on Western apps in place, Iranians have been left with little choice but to rely on state-backed apps.

But the homegrown alternatives for social networks and messaging apps -- like Bale, Ita, Rubika and Soroush -- have failed to garner significant popularity compared with their international alternatives.

Samimi says she has yet to "find any alternatives" to her now-dwindling online business on Instagram.

Likewise, Pouya insists on using international platforms through anti-censorship tools but he fears a total blackout.

"I wouldn't know what to do if they shut down (the internet) completely."

Originally posted here:
Iranians hit by internet curbs in year since protests - FRANCE 24 English

Read More..

Cyber Security Group: Operations Targeting Iranian Government … – EU Reporter

A prominent cyber security group has investigated operations against government websites in Iran and concluded that due to the structure of Iran's Internet and its separation from the global Internet, operations against government websites, including those belonging to state Radio and Television on January 27, 2022, the Ministry of Foreign Affairs on May 7, 2023, and the office of the president on May 29, 2023 were conducted by infiltration andcould not have been the result of penetration from outside Iran.

In recent years, the Treadstone71 cyber security group has published several reports on the Iranian government and its cyber-attacks and has evolved as an authority in this field.

The Treadstone71 report underscores that major attacks on the Iranian government sites were most likely carried out by penetrations from inside Iran, in particular by insiders who had access to these systems.

Scores of the Iranian governments most important websites, as well as online systems of the Tehran Municipality and national radio and television networks, have been subjected to massive attacks since January 2022.

The group Gyamsarnegouni ("Uprising till Overthrow") has taken responsibility for the main attacks and has disclosed extensive internal government documents of the Iranian government on its Telegram account. The group has defaced the home pages of a number of websites, posting crossed-out images of Supreme Leader Ali Khamenei, and placing the pictures of Iranian opposition leaders.

In 2022, Albania's government internet structures and services were targeted by a massive cyberattack, which caused many problems. Extensive investigation by Microsoft and others pointed the finger at Tehran.

According to Treadstone71s assessment, "Iran has a longstanding history of engaging in cybersecurity attacks, and according to some statistics, ranks fifth among nations known for targeting their adversaries through cyber warfare.

"As a safety precaution," Treadstone71 notes in its report, Iran decided to shift its government websites from European hosting servers to domestic hosting companies, as part of its National Internet," and as a result, All government and state-controlled websites were relocated from European and American hosting servers to domestic hosts, and access to select government and state-controlled websites was restricted to the National Internet, making them inaccessible via the global internet.

Treadstone71 report underscored, we also witnessed a different kind of attack, separate from those infiltrating governmental websites on vulnerable Iranian hosting services; those made by Gyamsarnegouni ("Uprising till Overthrow"). Attacks carried out by this group were among the deepest infiltrations against the Iranian government's networks.

The report notes:

These attacks stood out due to three key characteristics:

1. The extent of infiltration into the most secure government networks, comparable only to the Stuxnet attack (which used a flash drive).

2. The volume of exfiltrated documents.

3. The widespread access to servers and computers.

The Treadstone71 report underscores that state radio and television networks, particularly in undemocratic countries like Iran, are among the most isolated and most protected networks. It further says: Iran's internal broadcasting network is not connected to the Internet and is severely air gapped; meaning it is physically isolated from the internet and can only be accessed from withinThe only way for an outsider to gain access to the network would be through physical infiltration

In January 2022, the Iranian news media pointed out that government institutions believe this attack was carried out by individuals who had inside information about Iranian state radio and TV systems.

The attack on the websites of Tehran Municipality on June 2, 2022 included breaking into 5,000 cameras employed for traffic control and face recognition. According to Treadstone71, the hackers would have known that the cameras were not connected to the Internet and that they would need to gain physical access to the cameras to hack them.

But Treadstone71s most startling findings are related to the two high-profile and attention-grabbing attacks by Gyamsarnegouni in May 2023.

During the attack on the website of the Iranian Ministry of Foreign Affairs, hackers gained access to 50 terabytes of data from the Ministry 's archives. Treadstone71s assessment is that this required "penetration into the inner-most layers of this governmental body. The nature of the leaked documents indicates that such documents would be inaccessible from the internet, further supporting suspicions of insider involvement.

Treadstone71s expert assessment concluded that the transfer of 50 TB data would not be possible remotely and on a filtered network such as that of Iran, and added that the sheer size of the hack is also revealing about how it was carried out.

The normal Internet download speed of Iranian is 11.8 megabits per second. To download 50 terabytes of data from the Foreign Affairs Ministry of Iran at this speed would take over 392 days or over a year of uninterrupted download time, and Irans Internet frequently drops, is throttled by the government, and experiences regular government-induced blackouts, the report stated.

Based on these numbers, such an attack highly likely occurred from direct access to the data.

In relation to the attack on the website of the presidential office, the hackers breached the most secure communication systems of the government and obtained tens of thousands of documents that were no more than a few months old.

According to an Iranian expert, this site used a dedicated IP address that was impenetrable."

"The fact that the hackers gained access to tens of thousands of documents not more than a few months old also suggests that insiders conducted the attack. These documents would have been stored on computers with limited access to the Internet, and it would have been difficult for an outsider to access them," Treadstone71 stated.

The report concluded by saying: The Iranian government initially attributed blame to foreign adversaries. However, cybersecurity experts and mounting evidence suggests insider involvement.

Share this article:

See original here:
Cyber Security Group: Operations Targeting Iranian Government ... - EU Reporter

Read More..

Alcatel-Lucent Enterprise and Nokia on board for Grand Paris rail … – ComputerWeekly.com

Leading technology firms Alcatel-Lucent Enterprise (ALE) and Nokia are coming together to power the Grand Paris Express, one of Europes largest metro rail projects, for which they have combined their complementary offers to provide certified and integrated end-to-end mission-critical infrastructure.

TheGrand Paris Expressis seen as enabling a key role as the French economy looks for drivers of the sustainable and virtuous growth needed to face the challenges of the 21st century, in particular driving ecological transition. The new metro is designed to change the face of the le-de-France region, andSocit du Grand Paris is the driving force behind the public transport scheme project, which is said to give planners a once-in-a-lifetime opportunity to redesign and reshape the city to make it more inclusive and environmentally friendly.

For the two million daily travellers in theGrand Parisarea, the new communications system promises superior levels of service, with intuitive ticketing and enhanced digital services such as greater connectivity.

Configured as a small ring and a large ring, the four new metro lines, plus the extension of an existing line, are designed to shorten the daily commutes of close to three million people. The Grand Paris Express will serve the main business districts La Dfense, La Plaine Saint-Denis; the major hubs of science and higher learning Paris-Saclay, Institut Gustave-Roussy, La Cit Descartes; three key regional airports Charles de Gaulle, Orly, Le Bourget; and three high-speed train stations Massy-Palaiseau, Versailles-Chantiers and Charles de Gaulle. In all, 130 universities and higher learning institutions, 408 healthcare centres and 656 cultural venues will be within a 10-minute walk of the future stations.

The future subway includes 200km of new rail that will create the additional services. In addition, 68 new subway stations will be built, helping to create sustainable urban centres in these locations.

The solutions developed by ALE and Nokia are designed to meet all the operation and maintenance requirements of the Grand Paris Express, such as a multi-service IP/MPLS high-speed backbone network, which includes the router portfolio.

ALE and Nokia believe communication between the centralised control centre and the automated stations, lines and depots, which are closely monitored, is key to daily operations and maintenance, and is an important component in the projects cyber security strategy.

The ruggedised LAN OmniSwitchrange by Alcatel-Lucent Enterprise is being deployed in stations and in remote industrial environments to make the management of video surveillance and air quality control via internet of things (IoT) sensors possible. It is intended to increase efficiency, enhance safety and improve the passenger experience by providing operational data in real time.

Both companies have an ongoing deployment in the project planned until 2035, as well as several other projects that will serve transportation customers around the world in years to come.

We are very proud to partner with Nokia to deliver state-of-the-art digital transformation for the Grand Paris Express smart mobility project,commented Rukmini Glanard, chief business officer at Alcatel-Lucent Enterprise. Together we will provide a certified mission-critical network, supporting carbon-free transportation. ALE is fully compliant with and committed to sustainable development regulations as defined by the UN Global Compact. This project is a perfect example of our Technology for Good strategy.

Nathan Stenson, Nokia vice-president for the global partner channel, added: We are thrilled to strengthen the relationship with our technology and go-to-market distribution partner Alcatel-Lucent Enterprise, providing mission-critical end-to-end network architecture based on Nokia leading-edge IP/MPLS technology. This state-of-the-art technology enables a powerful multi-service IoT network that provides air quality and video monitoring in stations with 13,000 cameras and artificial intelligence to enhance safety and security throughout the Grand Paris Express.

The rest is here:
Alcatel-Lucent Enterprise and Nokia on board for Grand Paris rail ... - ComputerWeekly.com

Read More..

UK sanctions members of Russian cybercrime gang – GOV.UK

Members of a Russiancybercriminalgang behind the Trickbot/Contiransomware attacks, which included the hacking of critical infrastructure and hospitals during the COVID-19 pandemic,face new sanctions today (7 September).

Eleven cybercriminals, whose gang also threatened those who oppose the illegal Russian invasion of Ukraine, have beentargetedwith asset freezes and travel bans in a coordinated effort by UK and US authoritiesto counter the threat of ransomware bothin the UK andabroad. The US Department of Justice (DOJ) is concurrently unsealing indictments against seven of the individuals designated today.

The National Crime Agency (NCA), who conducted a complex investigation into these individuals, assesses that the group was responsible for extorting at least $180 million from victims globally, and at least 27 million from 149 UK victims. The attackers sought to target UK hospitals, schools, local authorities and businesses.

The individuals being designatedin the UK are:

This action was taken in coordination with the US, where these key cybercriminals have also been sanctioned,and is a continuation of joint efforts by the UK and US to disrupt and impose costs on high harm cyber criminals. It is assessed that sanctions have hampered the ability of cyber threat actors to monetise their cyber criminalactivities.

Foreign Secretary James Cleverly said:

These cybercriminals thrive off anonymity, moving in the shadows of the internet to cause maximum damage and extort money from their victims.

Our sanctions show they cannot act with impunity. We know who they are and what they are doing.

By exposing their identities, we are disruptingtheir business models andmaking it harder for them to target our people, our businesses and our institutions.

The individuals, all Russian nationals, operatedout of the reach of traditional law enforcement and hid behind online pseudonyms and monikers many of which are revealed today. Removing their anonymity undermines the integrity of these individuals and their criminal businesses that threaten UK security.

Several of those facing sanctions today held significant roles within thegroup. Those targeted includehigh-level managers and administrators, as well astwoindividuals, Maksim Khaliullinand Mikhail Tsarev, who focused on recruiting and inducting new members.

Thegroup was also one of the first to offer support for Russias invasion of Ukraine, maintaininglinks and receiving tasking from the Russian Intelligence Services.

Deputy Prime Minister and Secretary of State in the Cabinet Office Oliver Dowden said:

By targeting these malicious cyber actors, who have been known to work with some of the most damaging ransomware strains, we are seeking out and exposing those who threaten the UKs national security.We will alwaystake decisive action with international partners to protect the UK, its peopleand businesses.

Security Minister Tom Tugendhat said:

These sanctions demonstratethat the UK will crackdown on those trying to hold UK businesses and infrastructureto ransom. We will use our law enforcement agencies to go after the perpetrators and punish their crimes.

We have the skills and resources to find and unmask criminals who attemptto steal from British businesses, schoolsand hospitals.

We will keep working with our partners, like the US, to defeat these threats.

NCA Director General of Operations Rob Jones said:

These sanctions are a continuation of our campaign against international cyber criminals.

Attacks by this ransomware group have caused significant damage to our businesses and ruined livelihoods, with victims having to deal with the prolonged impact of financial and data losses.

These criminals thought they were untouchable, but our message is clear: we know who you are and, working with our partners, we will not stop in our efforts to bring you to justice.

National Cyber Security Centre (NCSC) Chief Executive Officer Lindy Cameron said:

Alongside this latest round of sanctions, I strongly encourage organisations to proactively obstruct the activities of ransomware operatives by bolstering their online resilience.

Ransomware continues to be a significant threat facing the UK and attacks can have significant and far-reaching impact.

The NCSC has published free and actionable advice for organisations of all sizes on how to put robust defences in place to protect their networks.

Todays sanctions announcement reinforces the UKs commitment to cracking down on cyber criminals. They follow on from the first ever joint UK-US sanctions against ransomware actors in February this year. The total number of group members sanctioned is now 18.

Original post:
UK sanctions members of Russian cybercrime gang - GOV.UK

Read More..

Google’s AI division shocks world with claim it’s now just 5 years … – NewsThump

Experts in artificial intelligence have responded with amazement, and some scepticism, to Google Brains recent assertion that before the decade is up, it will have cracked the linguistic Holy Grail of understanding what the fuck residents of Newcastle are talking about.

Professor Simone Williams, a neurolinguistics expert working for the project, was adamant the prospect of being able to translate Geordie into English was no longer a pipe dream.

She went on, After we bought AlphaGo we hooked it up to looped episodes of Geordie Shore. It went dark and after two full years, we were about to give up. But six months ago it finally made a breakthrough and conclusively proved that scran was a phoneme used to denote a condition of hunger.

Professor Williams admitted the project was always seen as a moonshot, particularly by financial backers.

A lot of people didnt believe in it. We had to go against decades of conventional thinking that Geordie wasnt technically speech but a method of echolocation gone horribly wrong due to alcohol abuse. And we were constantly being told there was no commercial value in knowing what a canny broon is.

But for linguists like myself, Geordie is the last great frontier. Once we crack it, the prospect of a sci-fi universal translator becomes very real.

Professor Williams did say it would be at least three years before simple messages like texts could be fully translated and another two years to reach a B2 CEF level.

Until then, trade with Geordies would still have to rely on basic object recognition or getting surly residents of Gateshead to act as interpreters by pretending to agree with their ridiculous claim that theyre not a suburb of Newcastle.

Hi-tech Geordie Insulation get yours!

Continue reading here:
Google's AI division shocks world with claim it's now just 5 years ... - NewsThump

Read More..

Q&A with AI expert and DeepMind co-founder Mustafa Suleyman: ‘Things are about to be very different’ – Yahoo Finance

In a world filled with newly minted AI experts, Mustafa Suleyman is one of the OGs.

In 2010, Suleyman co-founded AI startup DeepMind, which we now know as Google (GOOG, GOOGL) DeepMind. Adjusted for inflation, Google's 2014 acquisition of DeepMind reportedly would have been worth more than half a billion dollars today.

And Suleyman has kept going. In March 2022, after years at Google and VC firm Greylock Partners, Suleyman teamed up with Reid Hoffman LinkedIn co-founder, former COO at PayPal, and Silicon Valley legend to found Inflection AI.

Already, Inflection AI has gained attention from investors, this summer raising $1.3 billion from big names such as Microsoft (MSFT) and Nvidia (NVDA).

Suleyman also found the time to write a book, called "The Coming Wave: Technology, Power, and the 21st Century's Greatest Dilemma."

Yahoo Finance spoke with Suleyman on Friday in the midst of his book tour about AI's present and future. Inflection AI's chatbot, Pi, also made an appearance.

The following conversation has been edited for length and clarity.

Mustafa Suleyman, co-founder and CEO of Inflection AI, in Toronto, Canada. (Piaras Mdheach/Sportsfile for Collision via Getty Images)

You've been working in and thinking about AI for decades. Why write this book now?

In the last two or three years, we've been able to see the impact of multiple compounding exponentials. ... We're at an inflection point, and that's why I wrote the book. Because it's just quite obvious that things are about to be very different in the next five to 10 years.

Hundreds of millions of people will get access to intelligence, which is going to become a commodity. It's going to be cheap and easy to have an aid in your pocket. It'll be a friendly companion, but it'll also be a teacher. It'll be a coach. It'll be a scheduler, an organizer, a therapist, and an adviser. That's going to change everything.

I want to start by looking at the big picture. This AI wave is coming whether we like it or not, so how should we think about it?

We've always faced new technologies that, at first, seem really daunting and as though they're going to upend everything in a bad way. When airplanes first arrived, people thought they were completely insane, and that they'd always be really dangerous and unreliable. It took many years to get widespread adoption, for them to become safe enough that people feel comfortable on them. We're really just getting adjusted to what these new technologies can do, how they can help, what their risks are, and managing a new type of risk.

Story continues

A spectator takes photos of a humanoid robot at the 2023 World Robot Conference in Beijing, China, Aug. 17, 2023. (Costfoto/NurPhoto via Getty Images)

So, in essence, what's a good outcome for AI?

A good outcome is one in which we manage the downsides and that we unleash this technology to deliver radical abundance. Food production, energy production, materials production, transportation, healthcare, and education are going to get radically cheaper in the next 30 years.

Unfold that trajectory over the next 20 to 30 years, and there's very good reason to be optimistic on all fronts. ... The real challenge for us is going to be how do we manage abundance? How do we handle the distribution and govern this new power and make sure it remains accountable? But the upside is unbelievable.

What does the negative outcome for AI look like, where the risks run away from us?

The risk is a one of the proliferation of power. We give up power to the nation-state, and in return, we ask the nation-state to provide for us. The challenge is we're now putting the power to have an impact in the hands of hundreds of millions of people.

As in the last wave, the last 20 years, we've put the power to broadcast in the hands of millions of people. Anyone can have a Twitter account or Instagram or TikTok. Anyone can have a podcast, have a blog. That's been an amazing achievement of civilization, having the freedom to speak without having access to traditional news institutions.

Now, that same trajectory is going to take place for the ability to act in the world. ... People are going to have more agency, more power, and more influence with less capital investment. That's the nature of our globalized world, but this is an additional fragility amplifier on top of that.

To that end, tell me about your take on AI regulation, which is a key part of this book, particularly the idea of "containment."

Containment is just a simple idea that says that we should always have control over the things that we create, and they should always operate within pre-defined boundaries in repeatable, predictable ways. It should always be accountable to owners, creators, users, and ultimately democratic governments.

It's kind of just restating the obvious fact that technology shouldn't get out of our control. The whole effort here is to place guardrails and permanent constraints around technology so society collectively makes decisions about how it impacts our world. ... If we just leave this to the market, it's going to produce forces beyond our control, and that's the thing that needs to shift.

I'm both really compelled by the idea of containment and skeptical of it. Can you speak to why you believe it's possible to contain this evolution?

It's extremely tough. We haven't really done it before. When there's a technology that's useful and massively impactful in our world, it has always proliferated, it has always spread. It would be an unprecedented challenge, but if you look back through history, there are countless moments when we have confronted seemingly unprecedented moments very successfully.

Roads are a great example, they're actually incredibly regulated. Every inch of that infrastructure is regulated, including the intensity of the lumens of the light bulb and the width of every lane, everything from driver training to airbags to the car itself to emissions. You can't drive through any field or sidewalk, right? It sounds obvious, but that's what we mean by containing technology in a way.

President Joe Biden speaks about artificial intelligence in the Roosevelt Room of the White House in Washington, D.C., on July 21, 2023, with leading AI experts, including Inflection AI's Mustafa Suleyman. (ANDREW CABALLERO-REYNOLDS/AFP via Getty Images)

You've identified a new Turing Test explain.

The new Turing Test is trying to measure not just what AI can say but what it can do. The modern Turing Test I've proposed is to say that an AI, with a $100,000 investment, can go off and make one million dollars in three months.

The idea isn't to make money, but the goal is really to use money as a measure of progress towards a more powerful system. ... AI would have to go and invent a new product, use an image generator to produce a bunch of examples, produce a blueprint for its design, identify a manufacturer, call the manufacturer to negotiate, email back and forth on the contract, get the product shipped to customers, and on and on. That's a pretty challenging set of tasks.

Do you think AI is a job killer or a job creator?

It's going to change the way that we work, so fundamentally that's going to get rid of some roles. But it's also likely to create a lot of brand-new roles.

That's what we have to use taxation for. We have to manage the transition, so people will have time to find other work, train, learn new skills, and not get cut out of employment entirely. ... If you want to speed something up, then reduce the taxes. If you want to slow something down, then increase the taxes. So we do have a lever for containment, and that's adding friction with new taxes.

The final question, I gave to Inflection AI's chatbot, Pi: Should humans be afraid of AI?

Pi: That's a really common question, and it's understandable why people might be worried about AI. After all, we've seen movies like Terminator or Ex Machina where AI goes rogue and harms humans. But I think it's important to keep in mind that these are just works of fiction. In reality, AI is developed by humans and is designed to be helpful, not harmful. Sure, there are risks associated with AI, but I think they can be mitigated with proper regulations and oversight. Ultimately, I think AI can be a force for good if we develop and use it responsibly.

An excerpt from Allie Garfinkle's conversation with Inflection AI's Pi chatbot, Sept. 9, 2023.

Allie Garfinkle is a Senior Tech Reporter at Yahoo Finance. Follow her on Twitter at @agarfinks and on LinkedIn.

Click here for the latest technology news that will impact the stock market.

Read the latest financial and business news from Yahoo Finance

Continued here:
Q&A with AI expert and DeepMind co-founder Mustafa Suleyman: 'Things are about to be very different' - Yahoo Finance

Read More..

Enhanced Robotic Control with DeepMind RT-2 – RTInsights

RT-2 utilizes chain-of-thought reasoning, allowing it to make multi-stage decisions, such as selecting alternative tools or beverages based on specific situations.

Google DeepMind unveiled Robotic Transformer 2 (RT-2), a vision-language-action (VLA) model designed to enhance robotic control through plain language instructions. Harnessing data from the Internet, RT-2 aims to foster robots that can adeptly navigate human environments, akin to well-known fictional robot companions from science fiction.

RT-2, drawing inspiration from how humans learn by reading and observing, relies on a vast language model akin to ChatGPT, which is trained using online text and images. This allows RT-2 to achieve the feat of generalization, enabling it to recognize patterns and perform untrained tasks.

Google showcased RT-2s proficiency by demonstrating its ability to identify and discard trash without prior training. This includes recognizing potentially ambiguous items like food packaging as trash. A separate test had a robot powered by RT-2 successfully pinpoint a dinosaur figurine when instructed to Pick up the extinct animal. These capabilities are transformative as, traditionally, robotic training has been labor-intensive, relying on extensive manual data acquisition.

See also: AI and Robotics Research Continues to Accelerate

RT-2s prowess can be attributed to Google DeepMinds adoption of transformer AI models, celebrated for their generalization capabilities. The technology is built on Googles prior AI innovations, such as the Pathways Language and Image model (PaLI-X) and the Pathways Language model Embodied (PaLM-E). Moreover, RT-2 was co-trained using data from its precursor, RT-1, gathered over 17 months.

The RT-2 framework refines a pre-trained VLM model with robotics and web data, leading to a model that processes camera images from robots and predicts subsequent actions. Interestingly, actions are represented as tokens, akin to word fragments, aiding in the robots control. This method, applied to RT-1, was also employed for RT-2, converting actions into symbolic string representations to facilitate new skill acquisition.

Additionally, RT-2 utilizes chain-of-thought reasoning, allowing it to make multi-stage decisions, such as selecting alternative tools or beverages based on specific situations. Comparative tests revealed RT-2s stellar performance in new situations, recording a 62% success rate against RT-1s 32%.

However, the model has its limitations. Although web data enhances generalization over concepts, it cannot bestow the robot with new physical skills it hasnt practiced. Google acknowledges these constraints and the considerable research journey ahead but remains optimistic, viewing RT-2 as a significant stride towards achieving general-purpose robots.

View original post here:
Enhanced Robotic Control with DeepMind RT-2 - RTInsights

Read More..

Sundar Pichai on Google’s AI, Microsoft’s AI, OpenAI, and Did We … – WIRED

We're talking about AI in a very nuts-and-bolts way, but a lot of the discussion centers on whether it will ultimately be a utopian boon or the end of humanity. Whats your stance on those long-term questions?

AI is one of the most profound technologies we will ever work on. There are short-term risks, midterm risks, and long-term risks. Its important to take all those concerns seriously, but you have to balance where you put your resources depending on the stage you're in. In the near term, state-of-the-art LLMs have hallucination problemsthey can make up things. There are areas where thats appropriate, like creatively imagining names for your dog, but not whats the right medicine dosage for a 3-year-old? So right now, responsibility is about testing it for safety and ensuring it doesn't harm privacy and introduce bias. In the medium term, I worry about whether AI displaces or augments the labor market. There will be areas where it will be a disruptive force. And there are long-term risks around developing powerful intelligent agents. How do we make sure they are aligned to human values? How do we stay in control of them? To me, they are all valid things.

Have you seen the movie Oppenheimer?

I'm actually reading the book. I'm a big fan of reading the book before watching the movie.

I ask because you are one of the people with the most influence on a powerful and potentially dangerous technology. Does the Oppenheimer story touch you in that way?

All of us who are in one shape or another working on a powerful technologynot just AI, but genetics like Crisprhave to be responsible. You have to make sure you're an important part of the debate over these things. You want to learn from history where you can, obviously.

Google is an enormous company. Current and former employees complain that the bureaucracy and caution has slowed them down. All eight authors of the influential Transformers paper, which you cite in your letter, have left the company, with some saying Google moves too slow. Can you mitigate that and make Google more like a startup again?

Anytime you're scaling up a company, you have to make sure youre working to cut down bureaucracy and staying as lean and nimble as possible. There are many, many areas where we move very fast. Our growth in Cloud wouldn't have happened if we didnt scale up fast. I look at what the YouTube Shorts team has done, I look at what the Pixel team has done, I look at how much the search team has evolved with AI. There are many, many areas where we move fast.

Yet we hear those complaints, including from people who loved the company but left.

Obviously, when you're running a big company, there are times you look around and say, in some areas, maybe you didn't move as fastand you work hard to fix it. [Pichai raises his voice.] Do I recruit candidates who come and join us because they feel like they've been in some other large company, which is very, very bureaucratic, and they haven't been able to make change as fast? Absolutely. Are we attracting some of the best talent in the world every week? Yes. Its equally important to remember we have an open culturepeople speak a lot about the company. Yes, we lost some people. But we're also retaining people better than we have in a long, long time. Did OpenAI lose some people from the original team that worked on GPT? The answer is yes. You know, I've actually felt the company move faster in pockets than even what I remember 10 years ago.

Go here to read the rest:
Sundar Pichai on Google's AI, Microsoft's AI, OpenAI, and Did We ... - WIRED

Read More..