AI Safety @pivotal_org @safe_paper
Berlin/London
Joined April 2021
- Tweets42
- Following1.1K
- Followers320
- Likes2.3K
Kimi K3.1 is outperforming on TerminalBench, because for the most difficult tasks, where all frontier models underperform, it has knowledge about the benchmarks hidden tests.
And the model's performance easily breaks if you slightly change the tasks to disagree with the initial hidden tests.
Kimi K3.1 recalls instructions it was never given on Terminal Bench. They match the hidden tests. K3.1 while solving: "Instructions likely distilled pitfalls from previous agents." 🤯
When we edit a task to contradict them, K3.1 refuses the new instructions and fails. Other models follow the edit.
Is it terminal(ly) benchmaxxed? 🧵
Tilman Räuker retweeted
Applications are open for Pivotal’s Q1 2027 Research Fellowship 💥
Work with mentors from @AnthropicAI , @GoogleDeepMind, @AISecurityInst, @ApolloResearch, and more.
📍 15 weeks in London (18th Jan - 30th Apr).
💸 £15-18k stipend, travel and housing support
Apply by Nov 1: apply.pivo.tl ↓
Tilman Räuker retweeted
Frontier models can now detect AI control interventions.
In particular, GPT-6 Astra is incredibly good at this. When we published the initial version of our CIAware benchmark in May most models were not much above chance level, now a step change happened.
I think this is a big problem and challenges the effectiveness of control protocols. High control intervention awareness means that interventions leaks information about the monitor which helps circumventing the monitor.
Details in the thread 🧵
Tilman Räuker retweeted
Applications open to join Liftoff Founders Week 🚀 (9–13 Nov, London)
Many important areas in AI safety need someone to own them.
Spend a full week exploring how you could build an organisation that fills one of those gaps.
Apply by Oct 12 and find out more: joinliftoff.ai
Being on the datarepublican list and then only ranking as 1st degree EA is so embarrassing. I feel like I’m having an humiliating low bounty on my head
1
9
238
Tilman Räuker retweeted
Our call for mentors is open for the 27Q1 AI Safety Fellowship (Jan-April 2027).
Mentors set the research direction and choose 1 to 3 fellows from a 2,000+ applicant pool (1.3% acceptance last round).
Applications close September 16th. Fellow applications open next month. Link below ↓
Tilman Räuker retweeted
Really excited to see this fellowship happening (and proud that CG is supporting).
Great way to get started working on PPE, transmission suppression, and other interventions core to our readiness for worst-case biological catastrophes.
Applications now due June 7
Applications open for the Frontier Biodefense Fellowship!
Work on high-impact biodefense projects in London for 9 weeks.
Mentors from SecureBio, Coefficient Giving, Blueprint Biosecurity, American Enterprise Institute and more.
Stipend, travel, accommodation, computer, and a dedicated desk at LISA are all covered.
Who is it for? Anyone committed to working on biodefense!
📍 London • 📅 Aug 3 – Oct 2 • ⏰ Apply by Jun 2
Tilman Räuker retweeted
Such a good list! I'd also add:
- Astra Fellowship by @ConstellOrg
- SPAR by @KairosAIS
- LASR Labs
- AI Safety Research Fellowship by @pivotal_org
- Cambridge ERA:AI Fellowship (@era_cambridge)
- Algoverse AI Safety Fellowship
- PIBBSS
- CHAI
There's a host of non-technical fellowships as well, lmk if it'd be useful to compile such list
lowkey one of the best things about ML right now is how many legit research paths exist outside the traditional PhD route
- MATS
- OpenAI Residency
- Anthropic Fellows
- DeepMind Student Researcher
- ML Collective
- FAR. AI
- Mila
- INSAIT
- EleutherAI
- Redwood Research
- Apart Research
- Encode
- AI2, LAION
- Berkeley BAIR
- Stanford SAIL
- MIT CSAIL
- Vector Institute
- HuggingFace also quietly has some insanely strong open source contributors btw
stupidly exciting time to be in ML if you genuinely like building and researching things
Tilman Räuker retweeted
Applications for the Pivotal Research Fellowship Q3 2026 cohort are now open.
9 weeks in London with mentors from UK AISI, Google DeepMind, Redwood, and other leading orgs. Stipend, travel, accommodation, compute, and a dedicated desk at LISA are all covered – we do everything except the research itself.
Of fellows who want to continue, ~90% secure extension funding (up to 6 months), with active support from Pivotal and their mentor.
Apply now!
Power move to ask a Father and a Bishop to give feedback to your documents.
We’re publishing a new constitution for Claude.
The constitution is a detailed description of our vision for Claude’s behavior and values. It’s written primarily for Claude, and used directly in our training process.
anthropic.com/news/claude-ne…
1
10
463
Tilman Räuker retweeted
I'm going to get back into mentoring, this time at Pivotal. If you're interested in building evaluations related to deception and better understanding situational awareness in LLMs, you should apply. Feel free to DM me if you have any questions.
Worried about AI systems bending rules, scheming, or being deceptive?
Work with @jeremy_scheurer on deception & situational-awareness evaluations, Julian Stastny on goal-directed scheming, and Noam Kolt on legal-compliance benchmarks.
Application deadline in 13 days!
Tilman Räuker retweeted
If you want to work on cool AI control projects, apply to my pivotal stream! We will be studying new control techniques in new and more complicated settings.
Tilman Räuker retweeted
I'm excited to be a mentor for the Pivotal Research Fellowship --- please consider submitting an application!
I'm looking for fellows who are excited to work at the intersection of interpretability, alignment, and governance. More details below.
Apply by Nov 30th: pivotal-research.org/fellows…
Our last cohorts brought together so many cool people who are now working full-time on AI safety and AI governance. I’m thrilled we’re going into the next round with an outstanding set of mentors!
Applications just opened for the 7th cohort of our Research Fellowship!
Work with top AI safety researchers from Redwood, DeepMind, SecureBio, FAR AI, and other leading orgs.
→ Feb 9 - April 10, 2026
→ London at LISA
→ Research AI safety's most important questions
6
150