Uncategorized

DOJ Stands Up a Super Intelligence Task Force for AI Crime

US prosecutors now have a dedicated Super Intelligence Task Force. The Justice Department said on Thursday it will coordinate criminal and civil cases involving AI, absorb an earlier AI litigation unit, and work with the White House force that is listing the laws AI labs already face.

The new body is called the DOJ Super Intelligence Task Force, or DOJ-SITF. It is co-chaired by Associate Attorney General Stanley Woodward and Associate Deputy Attorney General Stephen Weiss, the Justice Department said in a press release. Attorney General Todd Blanche ordered it in an October 13 memo that became public with Thursday’s announcement.

“The Department of Justice will continue identifying and prosecuting individuals whose use of SI violates the law, including by maliciously targeting American taxpayers, critical U.S. infrastructure, or other aspects of our national security,” Blanche said. The department now uses “super intelligence,” or SI, for what most people still call AI, following a September presidential order.

What the task force will do

Blanche’s memo sets out a broad enforcement brief. The unit will organise the department’s criminal and civil AI cases, stop overlapping investigations, recommend what to prioritise, and write guidance on prosecuting unlawful AI-related conduct. It will also be the department’s main contact with the White House Super Intelligence Force led by Director of National Intelligence Jay Clayton, Nextgov reported.

Weiss said companies will be expected to keep “vigorous internal controls so that SI technology behaves as intended,” CoinDesk reported. The force is also meant to deter misuse by foreign states, transnational crime groups, and people or firms engaged in fraud.

It replaces an earlier AI unit

In January, then Attorney General Pam Bondi set up an Artificial Intelligence Task Force that focused more narrowly on civil litigation. Blanche’s memo folds that body into the new DOJ-SITF and transfers its work, staff and reporting duties. Components and US Attorneys’ offices are told to notify the new force when they open AI-related investigations or before they bring charges.

How it fits the wider policy push

The Justice Department move comes a day after Pentagon technology chief Emil Michael said the White House force would list every existing law that applies to AI labs within about two weeks. That list is not new regulation. The DOJ unit is about how those laws, and criminal statutes already on the books, get enforced against misuse of the technology.

Why it matters

A specialised DOJ unit does not create new AI rules on its own. It does signal that federal prosecutors intend to treat AI misuse as a coordinated national priority, from fraud and critical-infrastructure attacks to how companies police their own systems.

The post DOJ Stands Up a Super Intelligence Task Force for AI Crime appeared first on DailyAI.

DailyAI Briefing: Data centres lose 40 of 41 local votes, plus 4 more

Welcome to the DailyAI Briefing, the day’s five biggest AI stories. US voters turned down AI data centres in 40 of 41 local ballots, and three in four said they’d oppose one nearby. OpenAI’s agents found 948 ways to take over Linux at about $1,279 each, Washington promised a list of the laws AI labs already have to follow, Apple cut its staff’s daily AI budgets, and ChatGPT learned to make animated GIFs.

Let’s dig in.

Voters turn down AI data centres in 40 of 41 local ballots

40 out of 41. That’s how many local data centre measures went against the buildout on Tuesday, across 10 US states, according to Ballotpedia and results gathered by CNBC. Of the 41, 28 would limit data centres through bans, pauses or new approval rules, and 13 would have made building easier. Ohio held 21 of them. The one exception was Harwood, North Dakota, where voters rejected extra review of large industrial projects.

The NBC News exit poll found 76% of voters would oppose a new data centre in their own community, and 56% called AI data centres a crucial or major factor in their vote for Congress. Goldman Sachs chief economist Jan Hatzius estimates that states where Democrats are likely to add restrictions hold roughly 45% of planned US data centre capacity, Business Insider reported. The plain reading: the scarce thing for AI companies is no longer just chips or power, it’s a town willing to say yes. Our full story has the details.

OpenAI’s agents found 948 ways to take over Linux

$1,279 per flaw. That’s what OpenAI says it cost its AI agents to find and build a working attack for each of 948 separate Linux flaws that let an ordinary user take full control of a machine, in a research post published on Wednesday. The whole run cost $1.21 million. The agents listed 2,308 possible flaws, confirmed 1,724 in sealed-off test machines and wrote a fix for every one of the 948, The Hacker News reported.

These are local flaws, so an attacker still needs a foothold first, such as a stolen login. But bug-hunting contests have paid $20,000 to $70,000 for a single flaw like this, which makes the run look cheap. On Thursday evening OpenAI also named the model behind it, introducing GPT-6 Cyber for security teams. If finding serious bugs costs about a thousand dollars each, the bottleneck shifts to the human maintainers who have to review the fixes. Our deep dive breaks down every stage.

Washington to list the laws AI labs already face

535. That’s the number of members of Congress Emil Michael, the Pentagon’s chief technology officer, said he doesn’t want writing their own AI rules. Speaking at a Wall Street Journal conference in Napa on Wednesday, he said the government’s Super Intelligence Task Force would publish the existing laws and regulations that apply to AI labs “in the next week or two”, Reuters reported. The task force, set up in October and led by Director of National Intelligence Jay Clayton, reports to the President.

The timing isn’t subtle. Democrats have just won both chambers, and Texas Representative Greg Casar said AI bosses “should be called to testify under oath”, The Hill reported. A list of current laws isn’t new regulation, but it will set the starting point for whatever the new Congress tries. Our story covers who sits on the task force.

Apple cuts staff AI budgets from $300 a day

Some Apple engineers could once spend $300 a day on AI tools. That fell to $200 last month, some teams may get $100, and staff outside engineering could drop to as little as $25, Bloomberg reported. Staff use Claude, ChatGPT and Gemini, according to 9to5Mac, and some teams are being told to switch to cheaper or on-device models.

It’s a small line in a very big budget. Apple’s full-year results show research and development spending of $45.8 billion, up 32%. The point is the direction: after two years of telling staff to use AI as much as possible, big companies are starting to count the cost per task. We ran the numbers on what the caps mean per engineer.

ChatGPT now makes animated GIFs

4 billion images a week. That’s how many people now make with OpenAI’s tools, up from 3 billion in September, the company said as it launched Images 2.6. The headline feature is looping GIFs, made from a typed idea or an uploaded photo, on every ChatGPT plan. You can change the speed and edit frames, though, per OpenAI’s help page, you can’t edit a GIF you upload from elsewhere.

Animating your dog is fun. Animating someone else’s photo without asking is less so, and a moving picture of a real person is easy to pass off as genuine. It’s worth checking where a GIF came from before trusting it. Our story explains how it works.

Signing off

Today was about who gets to say no. Voters said it to data centres, Apple said it to its own staff’s AI bills, and Washington is trying to say it to Congress before Congress says it to the AI labs. Meanwhile, OpenAI showed how cheaply AI can now find the holes in the software the world runs on.

See you tomorrow.

The post DailyAI Briefing: Data centres lose 40 of 41 local votes, plus 4 more appeared first on DailyAI.

Voters Turned Down AI Data Centres in 40 of 41 Local Ballots

AI data centres were on 41 local ballots across 10 US states on Tuesday. Voters sided against them in 40. In an exit poll, three in four said they would oppose a new one in their own community.

Ballotpedia, which tracks local ballot measures, listed the 41 votes the day before the election. Of those, 28 would limit data centres through bans, pauses or new approval rules, and 13 would make building easier, mostly through rezoning or non-binding votes of support. Ohio held 21, more than half.

Voters backed the limits, or rejected the pro-building side, in every case but one, according to Ballotpedia and results gathered by CNBC. The exception was Harwood, North Dakota, where voters turned down extra review of large industrial projects, data centres included.

What people were voting on

A data centre is a large building full of computer chips. The ones built for AI need huge amounts of electricity, and many use water for cooling. Several of the measures set limits on exactly that. One in Kansas bans data centres that use more than 25 megawatts of power. Others ask for a public vote before any project above set land, water or power thresholds can go ahead.

Some margins were lopsided. In Wells, Maine, more than 80% of voters approved a 180-day pause that also cancels recent approvals. Six earlier votes this year all went the same way, Ballotpedia notes, including a ban in Monterey Park, California, that passed with 88.3%.

The number behind it

The NBC News exit poll asked voters whether they would favour or oppose a new data centre in their community. 76% said oppose and 24% said favour. 56% called AI data centres a crucial or major factor in their vote for Congress.

The issue reached the top of the ticket too. Democrats who campaigned on pausing or regulating new projects won the governor’s races in Texas and Ohio, CNBC reports. Ohio’s governor-elect, Amy Acton, backs a conditional pause until data centres cover 100% of their utility costs. Goldman Sachs chief economist Jan Hatzius wrote on Wednesday that states where Democrats are likely to add restrictions hold roughly 45% of planned US data centre capacity, with Texas alone about 25%.

Power bills are a big part of the anger. OpenAI has already promised to pay its own power costs, and the industry was losing public favour before a single vote was counted.

Why it matters

For people living near planned sites, a local vote has turned out to be a real way to stop or slow a project. For AI companies, the limit on growth is starting to look less like money or chips and more like permission from the towns where they want to build.

The post Voters Turned Down AI Data Centres in 40 of 41 Local Ballots appeared first on DailyAI.

Apple Cuts Staff AI Budgets From $300 a Day as Its Costs Climb

Some Apple engineers could once spend $300 a day on AI tools. Last month that fell to $200, and some teams may soon get $100. Staff outside engineering, many already capped at $100 a day, could drop to as little as $25.

The cuts were reported by Bloomberg on Wednesday, citing people familiar with the changes. Apple did not comment. Employees who need more can ask senior leaders for an exemption.

What the budget pays for

The limits cover tokens. A token is a small chunk of text or data that an AI model reads or writes, and providers such as OpenAI, Anthropic and Google charge by the token. Long jobs like writing and fixing code burn through a lot of them, and the most capable models cost the most.

Apple staff can use Claude, ChatGPT and Gemini, 9to5Mac reports. Some teams are now being told to switch to cheaper models, or to models that run on the device itself, where there is no per-token bill.

The number behind it

For scale, $300 a day over a typical year of about 250 working days comes to $75,000 per engineer, our own rough sum and a ceiling, not actual spend. A $100 cap brings that ceiling to $25,000.

Apple’s total spending is still rising fast. Its full-year results show research and development costs of $45.8 billion in the year to 26 September, up 32% from $34.6 billion. Apple has said the rise came mainly from infrastructure costs, including AI, and from headcount. Total operating expenses reached $75.5 billion, up 22%.

So the token caps are a small line in a very large budget. Bloomberg says it is unclear how much Apple will save, because many staff were not using their full allowance.

How staff are taking it

Reactions are mixed, according to Bloomberg. Some employees say the old $300 limit was more generous than anyone needed. Others read the cuts as a sign of wider cost cutting under new chief executive John Ternus and worry about layoffs. Some divisions have brought in outside consultants to review how they work.

Apple is not alone. Reuters, summarising the Bloomberg report, notes that Goldman Sachs recently restricted access to some AI coding tools after reviewing productivity and risk. AI bills can climb quickly when nobody is watching, as companies hit by stolen AI keys have found.

Why it matters

Companies spent the past two years telling staff to use AI as much as possible. Apple’s caps suggest the next phase is about cost per task, which will push workers toward cheaper and on-device models and make AI providers compete harder on price.

The post Apple Cuts Staff AI Budgets From $300 a Day as Its Costs Climb appeared first on DailyAI.

ChatGPT Now Makes Animated GIFs as Image Use Hits 4 Billion a Week

People now make 4 billion images a week with OpenAI’s tools, up from 3 billion in September. From this week, some of those images can move: ChatGPT will turn a typed idea or an uploaded photo into a looping animated GIF, on every ChatGPT plan.

The feature arrived on Wednesday with Images 2.6, OpenAI‘s new image model. OpenAI says it makes pictures 50% faster than the version before and keeps characters, products and brand details more consistent when you ask for changes. It is rolling out in ChatGPT on the web, desktop and phone apps, and in Codex, OpenAI’s coding tool, across all subscription tiers.

How the GIFs work

You ask for a looping GIF and describe what should move. OpenAI’s help page gives the example “Make a looping GIF of a paper boat bobbing on the water.” You can also upload a still photo and ask ChatGPT to animate it.

Once the GIF is made, you can change its speed, step through it frame by frame, save it or share it. Android Authority reports the speed control runs from half speed to double speed. To change the animation itself, you describe the edit in the chat. One limit: ChatGPT can’t edit an animated GIF you upload from elsewhere.

Before this, ChatGPT could not make a real animated GIF in one step, The Verge notes. OpenAI pitches it at marketers animating a mascot and people making reaction memes.

The number behind it

OpenAI says the 4 billion weekly images work out at more than half a billion a day across ChatGPT, Codex and the paid developer service. That is a rise of about a third since September. GIFs give people a new reason to keep generating.

For developers, OpenAI split the model in two. GPT-Image-2.6 Flare is the fast option and can produce a low-quality image in as little as six seconds. GPT-Image-2.6 Sunburst is the slower, higher-quality option. Both are paid by use.

What to watch

Turning a still photo into a moving one is fun with a pet or a doodle. It is less fun if someone does it to your photo without asking. A moving photo of a real person can be passed around as if it were genuine, and ChatGPT’s picture tools have already drawn scrutiny, from the “replica” image trend to fake election photos in the US midterms. OpenAI points to a system card, a document describing its safety tests, for the safeguards on Images 2.6.

Why it matters

A feature that once needed separate apps now sits inside a chatbot many people already use, on every plan. That will make animated AI images far more common in group chats and feeds, and makes it worth checking where a moving image came from before you trust it.

The post ChatGPT Now Makes Animated GIFs as Image Use Hits 4 Billion a Week appeared first on DailyAI.

OpenAI’s Agents Found 948 Ways to Take Over Linux, at $1,279 Each

OpenAI says a team of its AI agents found 948 separate flaws in Linux that let an ordinary user take full control of a computer, and built a working attack for every one. The run cost $1.21 million, or $1,279 per flaw. Bug-hunting contests have paid $20,000 to $70,000 for a single flaw of this kind.

The figures come from a research post OpenAI published on Wednesday. They describe one run in mid-October using an unreleased “cyber model”, a version of its AI built for security work. The company says it has already sent hundreds of reports and fixes to the people who maintain Linux, starting with the most serious.

Linux is the free operating system underneath Android phones, most cloud servers and the world’s supercomputers. Its core, the kernel, runs to more than 40 million lines of code. A flaw there can reach billions of devices.

What the agents actually found

Every one of the 948 flaws is what security people call a local privilege escalation. In plain terms: someone who already has a basic account on a machine, or has tricked their way onto it, can use the flaw to become “root”, the all-powerful administrator who can read every file and change anything.

That is serious, but it is not the same as a remote break-in. An attacker still needs a foothold first, such as a stolen login or a malicious app. The flaws turn a small break-in into a total one.

OpenAI ran the attacks against a standard desktop version of Linux with its normal protections switched on, not a weakened test machine. It used the same rules as Pwn2Own and TyphoonPWN, two contests where hackers are paid for working attacks.

How the numbers break down

The work ran as a production line of AI agents, each with one job.

First, “scout” agents picked out the parts of the kernel an ordinary user can reach. They chose 5,034 targets, about 2.8 million lines, or roughly 7% of the code. That step cost $3,426.

Next, “analyst” agents read those targets and listed 2,308 possible flaws within 12 hours, for $316,549.

Then “triage” agents tried to prove each one by actually triggering it in a sealed-off virtual machine. Within 30 hours they confirmed 1,724, about three in four, as The Hacker News also reported. OpenAI says only 3.1% of the original list came from the AI misreading the code.

Finally, exploit agents tried to turn the confirmed flaws into real attacks. They produced 1,169 working takeover attacks and 291 attacks that leak private memory. After removing duplicates that shared the same root cause, OpenAI counted 948 distinct takeover flaws and wrote a fix for each. Human researchers reviewed the results before anything went to the Linux maintainers.

A successful attack cost an average of $359 in computing to build. The biggest single bill was the attack-building stage, at $775,929.

The economics are the story

OpenAI applied the contests’ past prize money to its haul. At $20,000 to $70,000 per flaw, the 948 would be worth $18.9 million to $66.3 million on paper, about 16 to 55 times what the run cost. OpenAI calls its cost an upper bound and expects better tooling to bring it down.

A second test points the same way. Against kernelCTF, a Google reward programme that uses a stripped-down, hardened version of Linux, the system produced 21 takeover attacks for about $32,936. OpenAI estimates those would be worth around $1.5 million in rewards if accepted, roughly 45 times the cost. For comparison, it says kernelCTF has accepted 26 such attacks from everyone combined so far in 2026.

The models are also getting better quickly. In late July, an earlier OpenAI model, GPT-5.6 Sol, found 166 unique flaws of this kind for $513,690. The October run found 948, more than five times as many, even though the kernel had three more months of fixes in between.

OpenAI also says the agents used at least 10 attack techniques that outside kernel experts had not seen before. It has withheld the details for safety reasons.

The people who have to fix it

The flaws do not fix themselves. Linux is maintained largely by volunteers and company engineers who review every change by hand, and they were already under strain before this report.

At the Kernel Recipes conference this autumn, Greg Kroah-Hartman, one of the kernel’s most senior maintainers, said the project is now fixing 33 security issues a day, LWN reported. The number fixed per release has roughly quadrupled from a long-run rate of about 500, a jump he put down to AI tools. His message was “don’t panic”: the kernel’s processes are coping. But he also said the best models are still wrong about half the time, and described a batch of 79 AI-found kernel bugs in which only 15 were confirmed as real, and all 15 had already been fixed.

In August he went further for one corner of the kernel. In a post to the kernel mailing list, he said he would reject AI-written patches for the drivers/staging area, with an exception for tested security fixes, because “at least 1/3 of the results they generate are flat out wrong or harmful.”

OpenAI’s own post shows why that caution makes sense. In an early run, its system reported 200 successful takeover attacks. Human review found that 20 of them had reached root through a different, unknown flaw from the one they were assigned. OpenAI added an independent check afterwards.

The maintainers are moving fast regardless. OpenAI says that by the time its 10-day scan and review finished, 61 of the takeover flaws had already been fixed in the main Linux code, 43 had fixes waiting, and 299 were being discussed in public. OpenAI says many of those fixes came from other contributors working on the same bugs.

What it means for you

For most people the advice is unglamorous: install updates. Android phones, home routers, smart TVs and cloud services all run on Linux, and the fixes only help once they reach your device.

OpenAI’s own conclusion is blunter. It says that for now, a low-level account on a system that is behind on updates “should be considered functionally equivalent to root access for a well-resourced threat actor.” In other words, if someone capable gets any foothold on an unpatched machine, assume they can take all of it.

It is the latest in a run of AI security results. Last month AI helped uncover the PHP2Shell flaw, and Anthropic said Claude had proved 123 open-source libraries could not crash. OpenAI itself this week raised its bug bounty to $172,000 for the worst flaws in its own products.

Why it matters

The cost of finding a serious Linux flaw has dropped to about $1,300 of computing, well below what contests have paid for one. The same capability that OpenAI pointed at defence will, as OpenAI itself expects, reach attackers too, so the real test is whether volunteer maintainers and device makers can ship fixes as fast as AI can find the bugs.

The post OpenAI’s Agents Found 948 Ways to Take Over Linux, at $1,279 Each appeared first on DailyAI.