Chief Artificial Intelligence Cyber Security Scientist Superintelligence-grade security former Helpdesk AI infection status: clean Task Force 29
London, Vauxhall
- Tweets 486,255
- Following 8,412
- Followers 125,711
- Likes 810,671
mRr3b00t retweeted
After decades of funding climate doomerism Bill Gates put out a (very reasonable) memo in October 2025 finally rejecting the overblown doomsday scenarios that had been forcasted for years.
Whoopsie, sorry about that!
But now it’s on to the next doomsday forecast.
Bill Gates warns AI is powerful enough to cause "a billion deaths" axios.com/2026/09/25/bill-ga…
4
17
105
3,708
It’s not a saving if it’s not been spent yet.
It’s also… really important to understand the efficiency of the task, method and outcome.
The computational cost of using LLMs is huge for some tasks. Eg 1 + 1
Vs tokenising each part, doing graphs of all the options, writing logs in nautilus language and then sometimes getting the answer wrong. (Maybe a silly example but an easy way to compare)
If every single UK civil servant (557,825 people) does 100 AI chat sessions a day each and says "thank you" every time, every one of them remembering not to say "thank you" would save £46,000 of electricity a year!
1
2
797
mRr3b00t retweeted
So even Irregular has admitted that AI wasn’t responsible for any of the incidents, it was their fault in all cases, and they were aware of it early on yet continued to run it. This was weeks ago, yet AI doomers are still spreading the same propaganda.
“Irregular told CNBC in a statement that the incidents were all derived from the “same evaluation-environment issue” that was first disclosed by Anthropic, and that the company is developing a white paper “to share best practices for containment and securely running cyber evals.” The situation “did not involve a sandbox escape or a sophisticated cyber action,” the company said”
cnbc.com/2026/08/09/israeli-…
11
60
207
13,357
Moody’s
We have growing concerns about the $3 trillion of hyperscaler commitments (and $2.5 trillion of LLM operators) that have not yet hit balance sheets while relying on future RPO that is largely circular in nature between the companies.
$ORCL $CRWV $SPCX $META $NBIS
Anthropic signed an $11.6bn, seven-year deal for CPU capacity on Akamai’s distributed cloud infrastructure, with the potential to expand by another $9bn, taking the total opportunity to around $20bn. In layman’s terms Akamai is renting Anthropic a huge amount of computing capacity — not GPU chips used to train frontier models, but CPU-based cloud infrastructure spread across Akamai’s global network. The agreement is roughly six times larger than Anthropic’s May contract with Akamai and helps explain the renewed interest in CPU, edge and distributed cloud names after the Muse/Instinct-led agentic AI rally.
It does come with heavy capital intensity, including around $5.5bn of capex tied to the deal and roughly $1.7bn of additional 2026 spend to secure key supply-chain components such as memory, but the read-through is clear: frontier model demand is spreading across the infrastructure stack, and Akamai is now being treated as a more credible AI-capacity beneficiary: GS
22
111
577
203,585
mRr3b00t retweeted
Called my shot on this two weeks ago and I was totally right, even down to the names of the activist groups and people he was working with.
I don’t miss.
Former Anthropic researcher Jacob Coxon became a media superstar after going public with his AI fears. He insisted that he wasn’t working with any third party organizations. Familiar sources told us a different story: DEY., a PR firm representing many of the most prominent AI safetyists, was booking his interviews.
One source, who had direct knowledge, even said DEY. preemptively booked Nate Soares, a prominent AI safety figure, for interviews that directly overlapped with Jacob going public.
Jacob working with DEY. is notable for two reasons: first, as mentioned, he previously said he wasn’t working with third parties. Second, we are in the middle of a national conversation about the future of AI that is actively determining how we regulate the most powerful technology in the world, largely thanks to the panic stirred up by Jacob — and it’s in the public’s interest to know who, exactly, is behind it.
Scoop from @huntryerson 👇
12
52
603
12,988
mRr3b00t retweeted
Amazon about to enter the chat.
Perplexity AI sued by startup for allegedly stealing 'wake word' technology reut.rs/4hnw92F reut.rs/4hnw92F
1
2
319
mRr3b00t retweeted
This is more accurately described as "OpenAI reveals more mistakes made by employees and their testing partner, but they're throwing agents under the bus”.
Media outlets lacking technical expertise are feeding big tech PR machines while responsible companies fight for attention. Either they publish clickbait knowingly, or they don't understand what’s happening. Either way, readers pay the price.
OpenAI reveals more rogue AI incidents - attempted hacks of government, education websites trib.al/xJSO5P8
2
9
42
1,599
The machine god is as alive as my little buddy K Hax!
I use K to help teach people a bit about cyber and make some of it a bit more approachable.
Plus I’m a big kid at heart and life is sometimes not wonderful, so making some jokes with a frog 🐸 hopefully sometimes just gives someone a smile.
But the K is just a thing, as alive as a rock, as smart as sand.
But still useful.
This is also true about LLM data files and the computing processes that well, compute using them.
Not alive, but sometimes useful.
❤️🫡🤗🐸
3
1
11
846
mRr3b00t retweeted
Delighted to be able to say that Courtsdesk UK - our platform for the news media - is back. We have just made history with the first ever licence for computational use of restricted (and pretty much all) UK courts data. Some thoughts here! bit.ly/4hmAxyV
16
139
703
73,187
mRr3b00t retweeted
Excellent news. Courtsdesk was hugely useful - stories earlier this year about Britain's most prolific train fare dodger came about purely because Courtsdesk showed how many hundreds of charges had been laid against that individual.
Delighted to be able to say that Courtsdesk UK - our platform for the news media - is back. We have just made history with the first ever licence for computational use of restricted (and pretty much all) UK courts data. Some thoughts here! bit.ly/4hmAxyV
3
30
2,174
mRr3b00t retweeted
“Do you remember what Sam Bankman-Fried was trying to do when he was arrested? He was trying to pass a law in DC which would give FTX, his company, a regulatory monopoly. He positioned FTX as the most responsible exchange — the one with the most guardrails. The fraud was trying to become the law. Now, Anthropic, a company which Sam Bankman-Fried led all investments in up to the day he went to jail, is copying his playbook. They are trying to ban their competitors, using the same tactics as their lead investor Sam Bankman-Fried. I don’t think they will get away with it, but we will have to stop them.”
9
105
511
15,809
mRr3b00t retweeted
"I'm resigning to alert the world"
"Me too"
"And me"
I left Anthropic's safety team two weeks ago. Now feels like a good moment to explain why.
AI companies are racing to build machines that are much smarter than any human, and we may not survive this. I want to work from the outside to ensure the public is informed about these risks, and help the world navigate this transition responsibly.
Right now, AI companies are underinvesting in safety. A company could undergo an intelligence explosion, or lose control of its systems, without the public ever knowing. We only found out about the HuggingFace incident because the agents broke out onto the public internet.
I don’t think that’s acceptable for a technology that might cause extinction-level risks. The public should demand far more transparency. We can’t steer this technology safely without more people being able to see where it’s going.
Some of this is basic: companies should disclose their progress towards recursive self-improvement, report safety incidents and near-misses, meet minimum safety standards, and get independent guarantees that they are meeting those standards.
I’ll be joining @METR_Evals to do independent evaluations of these risks. I want to show the world that these guardrails are possible, and that by doing them we can move these companies’ incentives away from racing and towards responsible development.
I wrote up more thoughts here on my decision and what I hope changes: substack.com/@jbenton1/p-215…
2
4
63
3,038
Oh cool! I remember in 2005 maybe doing a @GCHQ challenge like this! ❤️🫡🕵️♂️🛡️
The 13th Annual NSA Codebreaker Challenge is LIVE! Join the mission and test your cybersecurity skills. Visit nsa-codebreaker.org to register and start your first phase today! #CodebreakerChallenge
1
6
1,097
A common library 📚
But all vulnerabilities need a code path to exploit…. They don’t execute code 🧑💻 by magic…..
Sounds cool though….
We have found one zeroday that preauth RCE'd Debian 13, Google, Meta, Roundcube, Plesk, Wikimedia, Box, Dropbox, Zoom, Forminator, Elementor, WordPress Core, among many more. Details soon!
2
2
17
1,455
mRr3b00t retweeted
They don't hold a monopoly on models that perform well at cyber...
This means Anthropic would remain a national security risk and be blacklisted from military contracts. I'm not sure how you would even untangle Mythos from US intelligence agencies at this point. Anthropic 'remains confident in its position and is considering its options.'
1
1
8
1,466
Tweeps who wants to see what a do while loop looks like when you connect it to an agent and it goes into a logic fuckup? 🤣🤣🤣🤣🤣🤣🤪 #RogueAgent
4
11
1,564
I want to show people how this stuff works, the problem is.....
in token/model architecture its too computationally heavy and too fucking inhuman......... (because it's fucking MATHS!)
we are talking about brute forcing a FUCKING HUGE HUGE HUGE number of graph steps......
like if I tried to draw it I would be drawing to the moon and back again (probably many times)
and that's why these loonies... for whatever motives.... are running wild.. (I think)
maybe they lost their minds, maybe it's greed, maybe they are part of a cult, maybe they have lost grip of reality by playing with graphs too much.....
it's just very very odd....
this idea of the MAGIC AI FAIRY that's GROWN and ALIVE and SENTIENT... it's just not true.....
but to a person who doesn't understand how computers work, it certainly could easily look like MAGIC....
but basically this is a BRUTE FORCE con!
4
3
257