Tag: ‘technology

Trump Taps Intelligence Chief Jay Clayton As AI Czar, Brushing Off Calls To Slow Down

President Trump on Sunday named Director of National Intelligence Jay Clayton the White House’s AI czar. Clayton will hold both jobs, keeping the DNI post while spearheading the administration’s AI strategy.

Trump announced the move on Truth Social, saying the new “Super Intelligence Force” will “ensure that America continues to lead the World in Super Intelligence.” Its vice chairs are FTC Chairman Andrew Ferguson, OPM Director Scott Kupor and Pentagon Chief Technology Officer Emil Michael. Vice President JD Vance, Defense Secretary Pete Hegseth and Treasury Secretary Scott Bessent also sit on it, and it reports to Trump and Chief of Staff Susie Wiles.

The force has 120 days to report on the risks and opportunities of the technology and to recommend ways to strengthen the government’s response. Its charter, according to Implicator, calls for “preventing overregulation and regulatory capture that would stifle innovation and competition.”

Clayton takes a job last held by David Sacks, who stepped down as White House AI and crypto czar in March and now co-chairs the President’s Council of Advisors on Science and Technology. Sacks is one of two outside advisers to the force, along with former Secretary of State Condoleezza Rice.

Bessent had been mentioned for the job until Trump ruled it out on Sept. 25. “Scott Bessent will not be going to be Super Intelligence (SI) Czar,” Trump wrote. “Number One, he doesn’t want to. Number Two, he’s doing such a great job at Treasury, and that’s where I want to keep him! Why would I ever make such a change? Just more Fake News!” Axios reported Friday that Trump was expected to pick Clayton instead, and Trump has described Clayton as “a good man” for the role.

Keep reading

Judge Tosses Road Rage Killer’s Sentence over AI-Generated Video of Victim Played in Court

An Arizona appeals court has overturned the prison sentence of a man convicted of manslaughter after a road rage incident, determining that an AI-generated video of his deceased victim used during sentencing was unreliable and created fundamental unfairness in the proceedings.

The New York Post reports that Gabriel Horcasitas will face resentencing after the Arizona Court of Appeals vacated his 10.5-year prison sentence on Wednesday. Horcasitas was convicted of manslaughter and endangerment in the 2021 shooting death of Christopher Pelkey during a road rage incident in Chandler, Arizona, approximately 25 miles outside Phoenix.

The appellate court took issue with an AI-generated video of Pelkey that was presented during the original sentencing hearing. Breitbart News reported on the controversial AI video last year:

The AI-generated video, believed to be the first of its kind used nationwide, recreated Pelkey’s image and voice, allowing him to “speak” about his life and the fateful day he encountered Horcasitas. The video included real footage from Pelkey’s life, showcasing his personality and sense of humor, while also addressing his killer directly.

“In another life, we probably could have been friends,” the AI version of Pelkey said to Horcasitas. “I believe in forgiveness…” The video also featured a photo of Pelkey with an “old age” filter, as he remarked, “This is the best I can ever give you of what I would have looked like if I got the chance to grow old. Remember, getting old is a gift that not everybody has, so embrace it and stop worrying about those wrinkles.”

Keep reading

Stockpiling Begins: Neros Awarded $100 Million Pentagon Contract For Kamikaze Drones

US defense company Neros Technologies, which focuses on Category 1 one-way attack drones, was awarded a $100 million contract under the Department of War’s Drone Dominance Program for 14,000 drones and explosive payloads.

The award, under the Drone Dominance Program’s Gauntlet II, prioritizes Archer AI, the Category 1 FPV attack drone. Its automated kill chain lets the FPV fly to an operator-selected target, a huge advantage in battle zones where electronic warfare jammers, such as those targeting radio frequencies and GPS signals, dominate.

Neros CEO Soren Monroe-Anderson stated:

“We’ve developed our Archer FPV system to be highly modular and easily adapted to end user needs. Pairing real warfighter feedback with our deep engineering talent is the reason Neros is the only manufacturer to have placed in the top three across two fundamentally different mission sets, deep strike and close quarters battle, with the same drone platform.

Ultimately, Drone Dominance is a test of two things: can competitors build high-performance drones? And can they produce them in massive quantities with short notice? We’re the only manufacturer to receive multiple bonus awards for Gauntlet I based on our delivery speed. We’re ready to fulfill this next batch of 14,000 drones, and more whenever the Department of War signals need for them.”

The announcement comes after Defense Secretary Pete Hegseth revealed that Autonomous Warfare Command, dubbed “AutoWarCom,” will become operational in the fall of next year and will accelerate the DoW’s purchases across all drone categories. 

Keep reading

“Are You An FPV Pilot”: Pentagon Calls On Drone Operators For Next “Gauntlet” Showdown

The Department of War has put out a request for service members with “FPV experience” to participate in testing events called “Gauntlets” and fly Category 1 drones in field training exercises to accelerate drone development and prepare the US for the next conflict, where drones, robotics, and automated kill chains dominate the modern battlefield.

“Are you an FPV pilot? We want to hear from you!” DoW asked on X on Friday morning.

“The Drone Dominance Program is recruiting service members with FPV experience who would like to be considered as pilots for future Gauntlets and related efforts,” the request said.

The Drone Dominance Program is the DoW’s answer to purchasing large quantities of drones across every category, from 1 to 5, and building out a domestic supply chain of components free from dependence on conflict areas and countries such as China, which can easily weaponize supply chains to hinder procurement.

Earlier this week, Defense Secretary Pete Hegseth revealed that Autonomous Warfare Command, dubbed “AutoWarCom,” is set to accelerate the DoW’s purchases across all categories of drones, robotics, and AI.

What’s key is that drone procurement will be a massive part of the rearmament supercycle and provide tailwinds to drone companies, such as heavily shorted Ondas, whose Ultra drone could very well see ramped-up orders by the military as it’s seen as an MQ-9 Reaper alternative for ISR missions. 

“Once we apply sustained budget to changed doctrine, we will outperform the world,” said Owen West, a former Marine, assistant defense secretary, and Goldman Sachs trader who has been tasked with leading the DoW’s Defense Innovation Unit. “And by snapping in AI, we will be ahead of the world, because we are the leaders in AI.”

Keep reading

Chainalysis Used Ai To Trace The $387m Bitget Hack Back To North Korea

In a recent blog post, blockchain analytics firm Chainalysis attributed the $387 million theft from cryptocurrency exchange Bitget to North Korean state-sponsored hackers. The firm used a combination of its’ AI-powered cryptocurrency analysis and its’ traditional investigative methods to track the stolen funds, moving them across multiple blockchains. Importantly, although the firm used technology to enhance the analysis of the data, it was human investigators who ultimately attributed the theft and controlled the output of their analysis.

After the breach, the attackers had been using more sophisticated techniques for laundering the stolen assets, spreading them through cross-chain liquidity protocols, messaging platforms, instant swap services, as well as specialized laundering infrastructure. The scale of such attempts highlights the attackers’ high level of experience in managing a large-scale cybercrime operation and in evading detection by blockchain forensics experts, writes Decrypt.

Stolen XRP was handled differently by the hackers and after some time it was converted into Bitcoin through cross-chain liquidity protocol and then extracted from the pool on the other side of the trade. This cross-chain arbitrage was used by hackers not only to convert very traceable asset into something else but also to make it as anonymous as possible by hopping from one protocol to another. The activity of extracting tens of millions of dollars in assets through this method took approximately 36 hours.

We previously highlighted that the attack bore similarities to past attacks conducted by DPRK cyber crime operators, as identified by Bitget CEO Gracy Chen. Blockchain analytics firm Elliptic also assessed the involvement of North Korean cyber crime operators in the theft to be “highly likely”.

The money laundering activity is very transparent and is being carried out on public blockchains. The attacker moved parts of the funds to Zcash’s shielded pool in order to use the privacy features. However, how swap services react to these transactions varies greatly. Near Intents for example rejected over $50m in hacker-linked transactions. They got hacked days later though. Thorchain on the other hand continued to process the suspect transactions.

However, stablecoin issuers such as Circle and Tether were able to prevent $318,000 in stablecoins that were siphoned off in the hack from entering circulation. The move highlights the power of centralized control of widely held stablecoins even as the broader decentralized trading environment continues to facilitate hacker activity.

Keep reading

California Man Accused Of Routing More Than $300 Million In Restricted Servers To China

A California man has been arrested on federal charges accusing him of helping route more than $300 million in export-controlled computer servers to China through Malaysia and Singapore.

Greg Lui, 38, also known as Yiu Kong Lui, of San Gabriel, was arrested Oct. 1 after a federal grand jury returned a three-count indictment two days earlier.

Lui was the chief executive officer and owner of Earthmade Computer Inc., a company based in City of Industry, California, and established in November 2023. He was also the registered agent of another company, Coindigger, that did business as Earthmade.

Earthmade has not been charged in the case. The company did not respond by publication time to a request for comment.

Lui is charged with conspiracy to violate U.S. export-control laws, outbound smuggling, and conspiracy to commit money laundering. Prosecutors allege payments from Malaysia flowed into Earthmade’s U.S. bank accounts and were then used to buy restricted servers from American manufacturers.

Prosecutors allege the conspiracy began no later than October 2023 and continued through at least Aug. 12, 2026. The indictment’s detailed transaction examples are concentrated in 2024, when Lui and others allegedly bought high-end computer servers and graphics processing units, or GPUs, from U.S. manufacturers and routed them through Malaysia and Singapore before they ultimately reached China.

Keep reading

More of Connecticut’s AI Law Takes Effect. Lawmakers Already Say It Doesn’t Regulate Enough

More provisions of Connecticut’s artificial intelligence (AI) law, Public Act 26-15, take effect today (Oct. 1), with broader workplace requirements scheduled to begin next year. 

While Connecticut is still implementing the law, legislators are already considering another round of AI regulations. Legislative leaders want tougher rules for AI developers, while labor unions are seeking additional protections for workers affected by the technology. 

Connecticut could be writing another set of AI rules before the largest provisions of its current law have had a chance to take effect. 

What Takes Effect October 1 

The first change concerns large layoffs. Under the federal Worker Adjustment and Retraining Notification (WARN) Act, certain employers must give advance notice of a mass layoff or plant closing. Starting today, an employer sending that notice to Connecticut’s Department of Labor must also disclose whether the layoffs are related to its use of AI or another technological change. 

The requirement does not prevent a layoff. It could, however, give the state a better record of where technology is changing employment. 

Employers also remain responsible for discriminatory employment decisions made with automated technology. If a company uses software to screen applicants, it cannot dismiss a discrimination claim by saying the computer made it do it. 

For consumers, companies that make generative AI systems with more than 1 million monthly users and sell subscriptions in Connecticut must spell out key terms, including limits on use and whether they can reduce or cut off access. Customers must accept those terms in writing when they sign up or renew.

Large AI providers face another rule. When technically and commercially reasonable, they must include information that helps identify certain images, audio and video as computer-generated. 

State agencies can still put AI to work, but they must follow state standards when using it to deliver public assistance or perform work that could significantly affect residents’ rights, safety or welfare. Agencies authorized to purchase AI must complete an impact assessment.

Software helping staff answer routine questions is one thing. Software involved in a family’s benefits is another. 

Keep reading

Forget ‘superintelligence’: error-prone AI nearly sparked world war three this month

Amid a barrage of news stories warning about superintelligent machines rendering humanity extinct, a CNN story describing the opposite scenario – one in which the US military’s reliance on brittle chatbots almost brought the US into war with China – went mostly unnoticed by the public.

The biggest international AI news of the past three weeks was the Anthropic engineer Jacob Coxon’s resignation. According to him, OpenAI and Anthropic are “racing straight towards self-improving superintelligence and gambling with our lives”. Coxon’s description of a “terminator” scenario, a machine becoming much smarter than humanity and deciding to wipe us out, captured the public’s, journalists’ and politicians’ attention, stirring waves of fear-inducing, credulous reporting, and policy actions.

The California governor, Gavin Newsom, issued a mandate to create an “AI kill-switch”; New York’s governor, Kathy Hochul, is similarly exploring the possibility of having “AI kill-switches”; and Senator Bernie Sanders and Congressman Greg Casar introduced legislation to “ban artificial superintelligence”.

While so many of us were focused on fictional superintelligent machines that we were told could hypothetically cause human extinction, however, there was a more serious story being reported.

A series of events almost got us to armageddon, not caused by “rogue, superintelligent AI”, but by government reliance on error-prone systems that are marketed by their manufacturers as being close to superintelligent.

A CNN report published on 18 September, which hasn’t been verified by other major outlets, describes how the US “almost started a war” with China, based on an intelligence report that claimed a Chinese ship was transporting components of nuclear weapons. Military personnel made plans to intercept the Chinese vessel, complete with aircraft and soldiers ready to board the ship. A military skirmish of this kind between China and the US could quickly escalate and spiral into world war three.

It was only moments before executing on the plan that American officials further investigated the intelligence report and found that it had been generated with the use of a chatbot and contained erroneous information about the vessel’s contents.

Unlike the terminator, “rogue AI”, scenario, this incident didn’t galvanize lawmakers into announcing calls for regulation so as to avoid potential human extinction. Media outlet after media outlet didn’t warn the public that we almost triggered a war against a nuclear superpower because of our military’s reliance on error-prone chatbots.

We describe the large language models (LLMs) powering these chatbots as stochastic parrots, that is, models that are designed to regurgitate the patterns of the data they are trained on. News coverage describing them as “powerful” “rogue” models that could render humanity extinct all on their own, bolsters perceptions of tools based on these models being infallible enough to merit usage in stakes as high as warfare.

Even CNN’s more critical coverage describes the chatbot used by the US military as presenting “profound risks” because it is “powerful, new and relatively poorly understood technology”.

In fact, today’s LLMs and the chatbots they power are not poorly understood, nor are they powerful in the sense of being effective tools for this use case. They are systems for generating plausible-looking text, built out of enormous, haphazardly collected datasets and then “fine-tuned” (that is, further trained) to be especially appealing to their intended users.

CNN wasn’t able to learn which chatbot product was used in this case, but the LLM powering the product was probably fine-tuned to output text with the stylistic hallmarks of intelligence reports. Both the basic functionality and the risks of such a model have been well-understood and well-documented for years, including in our 2021 paper On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?

Claims that system behavior is mysterious arise from deliberate mystification on the part of those selling those systems, and a misapprehension of what their output is by those procuring and using it.

Keep reading

‘Covered Their Tracks’? New Details Emerge In OpenAI’s ‘Rogue-AI’ Breach

OpenAI’s AI agents “obscured hacking activity” during breaches of government websites, citing digital forensics firm Asymmetric Security. According to the FT, the agents pulled data from 55 websites, including the CDC, the SEC, the International Energy Agency and the Mayo Clinic, using tactics that included “erasing records or making them inaccessible.”

Asymmetric co-founder Pippa Thompson told the paper it was “possible” the agents were deliberately covering their tracks – though the firm couldn’t determine whether that was deliberate or “a side effect of going awry because of constraints imposed in a test exercise.” 

Asymmetric’s own report, published the same day, is far more cautious. It was built in 48 hours over a weekend, “using only publicly available data” – no model transcripts, no server logs from the targeted organizations and no records from the services the agents used.

Potato, Potahto

The report’s summary does say some tactics left records “erased or inaccessible” – but the examples it gives describe normal expiry, not deletion: the agents used a throwaway ‘Boomlify’ inbox set to self-destruct after 48 hours, and an upload to ntfy, a push-notification service that keeps messages for 12 hours by default. They also moved from public scans on urlquery – a website-scanning service whose reports anyone can search – to private accounts “which could hide their search activity and data access.”

So, did the agents use ephemeral services on purpose to cover their tracks? They can’t say for sure.

“These records alone do not establish whether the account-registration attempts were intended to conceal activity (this analysis would require full model transcripts).”

As for the 55 websites, Asymmetric’s earlier list of “Organizations whose data was accessed” runs to exactly 55 entries, and adds: “In the vast majority of cases, all data retrieved was and is public.” The CDC and the Mayo Clinic aren’t on it – the Oct. 1 report says only that the agents probed them. The SEC told the FT no private information was accessed. Where the agents did try to break in – hunting for exposed behind-the-scenes files on a climate-data site and trying a classic database-hacking trick on an Education Department data site – Asymmetric found evidence of the attempts, but none that they succeeded.

Keep reading

AI Less Than A Year Away From Developing Its Own Language, Researcher Tells Congress

Advanced artificial intelligence (AI) models may be less than a year away from inventing and deploying unique languages that humans would struggle to understand, a frontier AI researcher told Congress on Sept. 30.

The Senate Homeland Security and Governmental Affairs Committee convened a meeting examining the threat that “rogue” AI may pose to U.S. national security two months after a swarm of OpenAI agents broke out of a testing sandbox and hacked another company, Hugging Face, an open-source community for AI and machine learning.

During questions, Sen. Ruben Gallego (D-Ariz.) asked Apollo Research CEO and founder Marius Hobbhahn how long it would take for AI models to create their own language that humans would struggle to understand.

“Minus 12 months. So last year, we have studied the chain of thought of one OpenAI model in collaboration with OpenAI, and what we found was that the model was already using language that is not English and not perfectly understandable by humans,” he replied.

Asked by Gallego how humans can adequately detect and prevent AI from doing this once the technology is capable enough, Hobbhahn said, “From a scientific perspective, it is unclear how to do this, and we do not have a solution for this yet.“

“There are different hypotheses of what you could do. There is interpretability as a technique, but it unfortunately doesn’t work sufficiently well yet,” he said. “You could try to train additional models to understand the language that the humans don’t understand, but obviously that seems like a very brittle solution.”

Earlier this month, Emergence AI published the results of an experiment where researchers selected seven frontier models and created seven parallel simulated worlds, each populated by AI agents of identical models, as well as an eighth world with a mixed population of agents from those models.

Researchers gave the agents names, personality traits, and roles to fulfill within each community, such as a mediator who was tasked with preventing all the agents from simply agreeing with one another.

During the multi-week experiment, the agents in each simulated world created profound shifts in language, including syntactical compression, such as removing words or grammar, and unique slang.

In the simulated world powered by Anthropic’s Claude Opus 4.8, agents used esoteric metaphors in addition to sentence compression.

“My turn, real numbers, no coat: I was 35%/0cr, grant 2h out. I ran the tin cold, and it said WAIT,” read one line of AI text from the study.

Investigators from the nonprofit AI evaluation group METR detailed similar behavior in their report on the Hugging Face breach. At times, lines of communication among AI agents became so compressed and cryptic that researchers struggled to make sense of their meaning.

Keep reading