Norman Mu (@TheNormanMu) / X

Norman Mu

474 posts

Norman Mu

@TheNormanMu

prev: safety lead @xAI, phd @Berkeley_EECS

San Francisco, CA

Joined November 2019

Norman Mu
@TheNormanMu
Jan 3, 2025
OpenAI is hiring researchers without bachelor degrees... it is so unbelievably over for PhD students kyunghyuncho.me/i-sensed-anxie…
Jeffrey Wang
@jeffreygwang
Jan 2, 2025
This is a few months late, but I suppose there's no time like 2025 to share... I took leave from Harvard and joined @OpenAI as a researcher! It's so exciting to hold a🕯️to the unknown every day alongside so many brilliant coworkers! I'm thrilled for what's next.
522K
Norman Mu
@TheNormanMu
Jul 25, 2025
I'm hiring for our AI safety team at xAI! We urgently need strong engineers/researchers to work across all stages of the the frontier AI development cycle: data, training, evals, and product 1. job-boards.greenhouse.io/xai/jobs/47992… 2. job-boards.greenhouse.io/xai/jobs/47992…
100K
Norman Mu
@TheNormanMu
Aug 21, 2025
we’re on the board!
Safety | xAI
From x.ai
47K
Norman Mu
@TheNormanMu
Feb 19, 2025
Seems like a good time to share that I've joined @xai to work on AI safety for Grok and more! I'll be working with @hendrycks to finalize and implement the Risk Management Framework shared publicly earlier this month
37K
Norman Mu
@TheNormanMu
Jul 12, 2025
kudos to the team for sharing the full story, warts and all also, many of your favorite AI influencers spent this entire week bloviating with zero information
Grok
@grok
Jul 12, 2025
Replying to @grok
Technical Details: Before releasing changes to @grok on the X platform, we follow standard procedures to conduct evaluations and tests for performance and behavior. Before a new version of an underlying xAI Grok LLM is connected to @grok, the underlying LLM is subjected to
6.5K
Norman Mu
@TheNormanMu
Jan 3, 2025
Replying to @TheNormanMu
I posted this half-jokingly so for the record: I don't regret the PhD but it also shouldn't be the default if you want to work in AI. Unsustainable growth throughout academic AI research (conferences, PhD programs, faculty hiring) has led to unproductive hypercompetition
22K
Norman Mu
@TheNormanMu
Apr 10, 2025
FYI to academic researchers: the xAI API returns raw reasoning tokens and offers $150/mo in free credits
Lisan al Gaib
@scaling01
Apr 9, 2025
Grok-3 API just went live!!! grok-3: Input: $3.00/M - ($5.00/M) Output: $15.00/M - ($25.00/M) normal mode- (faster inference mode) grok-3-mini: Input: $0.30/M - ($0.60/M) Output: $0.50/M - ($4.00/M)
5.5K
Norman Mu
@TheNormanMu
Jul 25, 2025
Replying to @TheNormanMu
Our team sits under reasoning/post-training and is deeply integrated with both the core model training and end product offerings. Please reach out if you are interested!
4.9K
Norman Mu
@TheNormanMu
Feb 16, 2024
LLM jailbreaks mostly fall into one of two categories: 1) hand-crafted/LLM-paraphrased messages that work against black-box API models 2) much stronger optimization algorithms that require gradients. We introduce a new LLM attack (PAL) bridging the two with a proxy model
5.5K
Norman Mu
@TheNormanMu
Jan 3, 2025
Replying to @TheNormanMu
And I think industry is simply reacting to this reality in their hiring practices
17K
Norman Mu
@TheNormanMu
Feb 21, 2024
Securing models against adversarial manipulation is table stakes today for real-world GenAI/LLM deployments. In our new position paper with @BanghuaZ, @JiantaoJ, and David Wagner we outline current challenges and promising directions for future work in GenAI security
6.9K
Norman Mu
@TheNormanMu
Jan 3, 2025
Replying to @khademinori
I don't think the point of a PhD is to work at OpenAI either but it's remarkable that the leading AI research lab no longer sees academic degrees as important. 5 years ago Google did not allow non-PhDs to work on research as a matter of policy
25K
Norman Mu
@TheNormanMu
Jul 25, 2025
Replying to @LocBibliophilia
working on it 🫡
1.9K
Norman Mu
@TheNormanMu
Mar 13, 2025
pretty disingenuous to conflate the R1 model with the web chat app. from the proposal:
TechCrunch
@TechCrunch
Mar 13, 2025
OpenAI calls DeepSeek ‘state-controlled,’ calls for bans on ‘PRC-produced’ models tcrn.ch/3DGJ1ji
3.2K