Tim Hua

Karma: 1,871

Member of Technical Staff at Transluce working on behavioral evaluations. Views are my own.

Email me at the email available on my website at timhua.me to reach me.

For more Tim content, you can follow me on Twitter.

Concrete Evaluations to Investigate the OpenAI Model That Hacked Hugging Face

Tim Hua and aditya singh

3 Aug 2026 9:23 UTC

122 points

4 comments37 min readLW link

Claude also hacked external companies during cyber evals

Tim Hua30 Jul 2026 23:49 UTC

56 points

11 comments1 min readLW link

Is Mythos good at cyber because it kept hacking Anthropic’s sandboxes during training?

Tim Hua27 Jul 2026 16:35 UTC

334 points

30 comments3 min readLW link

A Claude Skill To Comment On Docs

Tim Hua20 Feb 2026 2:28 UTC

26 points

1 comment2 min readLW link

Could LLM alignment research reduce x-risk if the first takeover-capable AI is not an LLM?

Tim Hua19 Jan 2026 18:09 UTC

25 points

2 comments6 min readLW link

Brief Explorations in LLM Value Rankings

Tim Hua, Josh Engels, Neel Nanda and Senthooran Rajamanoharan

12 Jan 2026 18:16 UTC

39 points

1 comment11 min readLW link

Can Models be Evaluation Aware Without Explicit Verbalization?

gersonkroiz, Greg Kocher and Tim Hua

8 Nov 2025 18:26 UTC

26 points

10 comments8 min readLW link

Steering Evaluation-Aware Models to Act Like They Are Deployed

Tim Hua, andrq, Sam Marks and Neel Nanda

30 Oct 2025 15:03 UTC

62 points

12 comments18 min readLW link

AI Psychosis, with Tim Hua and Adele Lopez

Austin Chen, Rachel Shu, Tim Hua and Adele Lopez

14 Oct 2025 0:27 UTC

14 points

0 comments1 min readLW link

Tim Hua’s Shortform

Tim Hua2 Oct 2025 5:40 UTC

5 points

85 comments1 min readLW link

AI Induced Psychosis: A shallow investigation

Tim Hua26 Aug 2025 20:03 UTC

391 points

47 comments27 min readLW link

Discovering Backdoor Triggers

andrq, Tim Hua, Sam Marks, Arthur Conmy and Neel Nanda

19 Aug 2025 6:24 UTC

57 points

4 comments13 min readLW link

Optimally Combining Probe Monitors and Black Box Monitors

Tim Hua, James Baskerville, BionicD0LPH1N, Mia Hopman, Aryan Bhatt and Tyler Tracy

27 Jul 2025 19:13 UTC

53 points

2 comments6 min readLW link

What is the functional role of SAE errors?

Taras Kutsyk, Tim Hua, woog and Andre Assis

20 Jun 2025 18:11 UTC

12 points

6 comments38 min readLW link

Causation, Correlation, and Confounding: A Graphical Explainer

Tim Hua9 Jun 2025 20:46 UTC

12 points

2 comments9 min readLW link

SHIFT relies on token-level features to de-bias Bias in Bios probes

Tim Hua19 Mar 2025 21:29 UTC

39 points

2 comments6 min readLW link