Joined June 2008
- Tweets 66,076
- Following 2,743
- Followers 45,043
- Likes 41,891
903 Photos and videos
Pinned Tweet
4 Nov 2019
I will have to print & frame this tweet :-))
4 Nov 2019
If you watch one talk this month, make it this one. Profoundly insightful on the complexity threat to security by @halvarflake given at NATOs CYCON.
Video:
err.ee/836236/video-google-0…
And slides:
docs.google.com/presentation…
8
18
242
Halvar Flake retweeted
This is a good post about how involved it is to rewrite load-bearing functionality in a memory-safe language:
swift.org/blog/migrating-tru…
Bonus points for a 13% performance improvement!
6
27
3,092
Halvar Flake retweeted
The government should not be regulating AI to this extent. Not like this.
I’ve been against onerous regs when Anthropic and the safety community was pushing for it.
And I’m against it now that they got what they asked for.
30
25
290
17,089
Halvar Flake retweeted
Jun 13
IDK I didn't notice any difference, Fable 5 is working just as well as it always has for me
Jun 13
As a result of a US government directive, we are suspending access to Claude Fable 5 for all users. You can continue to use all other Claude models.
Here’s what this means for you:
Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error.
On the Claude Platform, requests to Fable 5 will also return an error. Please update your integrations to other Claude models.
We know this is a disruption to your workflows; we appreciate your patience and support.
7
5
90
12,453
Jun 12
The fact that I can't touch my academic research from 18 years ago with Fable without triggering a model downgrade due to "cyber" is ... baffling.
11
13
277
12,480
Jun 12
Making this work will imply a revolution in CAD.
Jun 11
JEFF BEZOS JUST EMERGED FROM STEALTH WITH A $41 BILLION AI STARTUP CALLED PROMETHEUS
$12 billion raised. Valued at $41 billion. Coming out of stealth today.
The backers: Bezos personally, JPMorgan, BlackRock, Goldman Sachs, DST Global, and Arch Venture Partners.
The mission: do for engineering and manufacturing what large language models did for text.
Bezos is calling it an "artificial general engineer." Instead of training on words from the internet, Prometheus ingests data from the physical world to accelerate the manufacturing of skyscrapers, smartphones, jet engines, and everything in between.
In Bezos' own words: "Something that today was going to take 100 engineers 10 years to build, if you can change that to taking 10 engineers one year to build, you're just going to get way more things built."
This is Bezos' first CEO role since stepping down from Amazon in 2021. He's co-leading it with Vik Bajaj, former Google X executive.
(Source Semafor)
2
15
7,465
Jun 12
Fable just downgraded to Opus because I am calculating a Groebner base on one round of the block cipher PRESENT.
This is absolutely ridiculous. I can essentially not use Fable to review my 2008 MSc thesis without triggering "cyber safeguards".
A friend of mine had the down...
16
10
164
11,841
Jun 12
...grade happen when they asked whether PTRACE allows catching all signals in the traced process, as they are writing a profiler.
4
35
1,671
Jun 11
Invisible safeguards is such a nice word for deceit & sabotage :-)
Jun 11
We’re rolling out changes to make Fable 5’s safeguards for frontier LLM development visible.
Starting this week, flagged requests will visibly fall back to Opus 4.8—the same as our safeguards for cyber and bio. You will see this every time it happens. On the API, any flagged requests will return a reason for their refusal (coming to server-side fallback in the next few days).
We wanted to deploy Fable 5 to our users quickly and safely. Visible safeguards can be probed, so they have to be robust, which takes time to get right. Invisible safeguards can be targeted more narrowly, allowing us to ship quickly with very few false positives. We went with invisible safeguards for this reason—and that was the wrong tradeoff. You should have visibility into the safeguards we have in place, and why. We’re sorry for not getting the balance right.
Making the safeguards visible makes them easier to work around, so keeping them robust to jailbreaks will unfortunately mean more false positives while we improve the classifiers. We're also tuning our bio and cyber classifiers to trigger less often on harmless requests. We know this is frustrating and we’ll do our best to keep this period as short as possible.
If you think a request has been mistakenly flagged: run /feedback in Claude Code, click thumbs-down on the fallback in Claude.ai or Cowork, or file the safeguard appeal form for API requests. Your reports help us tune these classifiers and we appreciate your feedback.
support.claude.com/en/articl…
5
9
110
7,778
Halvar Flake retweeted
Jun 11
I don't think people in Europe (and the UK) are taking our technological (and therefore economic) divergence seriously enough.
A few disparate datapoints:
1. Our compute is woefully behind; three American labs each operate more AI compute than all of Europe combined
2. OpenAI has paused Stargate UK (indefinitely); our energy costs and regulatory environment are actively driving frontier infrastructure away
3. Mistral reportedly considering acquisition by SpaceX; Europe’s most valuable AI company is struggling to get the necessary resources to compete
4. FluidStack cancelled plans to build in France and moved HQ from London to the US; a company founded in the UK, that signed an MOU with the French government, chose American capital and contracts
5. Project Glasswing launched as a coalition of US firms - the most powerful AI model ever built was shared with Americans first and Europeans are still negotiating access
6. A Trump executive order gives the US government up to 30 days of exclusive federal access before a model's public release, and a say in which 'trusted partners' can use it first (American strategic interests are being baked into the architecture of who gets access to frontier AI, and when)
Those who wrote Europe 2031 are some of the few people taking this seriously. Well worth a read.
Jun 11
Here's a project I've been working on recently: a vision of what happens if Europe doesn't take AI seriously, inspired by AI 2027 europe2031.ai/
22
79
538
55,992
Jun 11
It's good that they are walking back. Publishing a postmortem what sort of catastrophic internal ethics failure led to assuming this was ok to try might also be a good trust-building move.
Jun 11
Very pleased to hear Anthropic have walked back this policy simonwillison.net/2026/Jun/1…
ALT “We’re changing Fable 5’s safeguards for frontier LLM development to make them visible.” Anthropic said in a statement to WIRED. “We made the wrong tradeoff and we apologize for not getting the balance right.”
5
6
35
4,933
Jun 11
Uh. Distilling from a model with guardrails somehow equips the trained models to learn things the teacher refuses to teach?
5
3
33
6,784
Halvar Flake retweeted
Jun 9
I’ll be honest that it would have been much more difficult to defend Anthropic against the DoW incursion had that incident occurred after this one. This is the company literally telling their customers, “we reserve the right to silently sabotage you.” I’d still have defended them, because the government trying to destroy a firm is still wrong, but man would it have been a harder case to make.
18
28
592
128,163
Halvar Flake retweeted
If you need me I’m over on instagram watching a Georgian dance ensemble boogie to Future
117
1,394
10,166
476,619
Halvar Flake retweeted
Jun 10
Idea: a sandbagging eval that generates interesting frontier LLM ideas and then tries to implement them with Fable-5, then measures rate of bugs with GPT-5.5. The ideas with the highest rates of bugs correspond to Anthropic's secret sauce that they don't want replicated
7
12
149
7,535
Halvar Flake retweeted
Jun 10
Imagine building a computer and not allowing its use in CS research. Thats some dystopian shit.
43
116
2,169
90,358
Halvar Flake retweeted
Jun 10
Everyone, please join Project Tapestry
thealliance.ai/projects/tape…
45
162
1,116
430,527
Halvar Flake retweeted
Jun 10
the art of technical writing is to appease both the p99 domain-expert and the curious p50
it's like a Pixar movie that speaks to child & parent
16
49
696
26,860
Halvar Flake retweeted
Jun 10
We consume data we did not create.
We inherit tools we did not invent.
We run on chips we did not make.
But when the commons bears fruit, we fence it.
Jun 9
mythos will be bad ON PURPOSE on ai "frontier llm research" tasks, this is very very sad for the research community
also the fact that this is un purpose not visible to the user is crazy
19
90
1,114
37,619
Halvar Flake retweeted
Jun 9
btw, we publish everything you need to build our Nemotron models including the recipes and pipelines directly.
github.com/NVIDIA-NeMo/Nemot…
Jun 9
mythos will be bad ON PURPOSE on ai "frontier llm research" tasks, this is very very sad for the research community
also the fact that this is un purpose not visible to the user is crazy
23
59
616
24,400
Jun 10
Adding safeguards against people doing work on frontier models? Seriously?
Jun 9
this is the biggest wake-up call to protect and nourish open source AI
if you don't build out sovereign and independent models infra closed labs will patronize you to an insulting degree
3
21
3,344