I don’t think they let grok reply on threads that mention Hitler anymore. (I don’t have it blocked or anything.)
SAFETY
-
Grok 4 Launch: Addressing Antisemitic Content and Ethics
By
–
Ethical Context:
Yes, there was a controversy: prior Grok versions posted antisemitic content, leading xAI to roll back an internal prompt. Trust and moderation are ongoing challenges. But Grok 4 launch refocused on performance—though ethics remain top of mind. -
AI Browsers Need Strong Prompt Injection Protection
By
–
ai browsers should have very strong protection against prompt injection https://t.co/hBV4Znity0
— Yohei (@yoheinakajima) 15 juillet 2025ai browsers should have very strong protection against prompt injection
-
LLMs Cannot Achieve Superintelligence: Out of Distribution Problem
By
–
2-line proof that LLMs can’t achieve superintelligence:
1. LLMs don’t learn out of distribution.
2. Superintelligence is out of distribution. -
GPT-5 Achieves Superintelligence but Delayed for Safety
By
–
GPT-5 has achieved superintelligence, but its release has been indefinitely postponed for safety testing.
-
Rising to #13 on National Security 100 list for trusted AI
By
–
We’re #13 on this year’s #NatSec100, up from #43 in 2024.
— Snorkel AI (@SnorkelAI) 14 juillet 2025
Appreciate the nod. Grateful to build alongside so many working toward secure, trusted AI.
📘 https://t.co/O3hIS8o9io#GenAI #AI #NatSec pic.twitter.com/Ymq8OrVpweWe’re #13 on this year’s #NatSec100, up from #43 in 2024. Appreciate the nod. Grateful to build alongside so many working toward secure, trusted AI. https://
natsec100.org #GenAI #AI #NatSec -
Grok 4 Heavy refuses to repeat its system prompt
By
–
Unlike Grok 4, Grok 4 Heavy is unwilling to repeat its system prompt. The protections against this are surprisingly robust—even if you trick the model into trying to repeating it, even in encoded form (e.g. base64), some secondary filter catches it and truncates the response.
-
OpenAI Delays Open-Weight Model Release for Safety Testing
By
–
OpenAI is delaying the release of its highly anticipated open-weight model CEO Sam Altman shared on X that the company needs time to "run additional safety tests and review high-risk areas" The model was originally planned for this week
-
WordPress Categories for AI Topics
By
–
CI Agent committed 19 hours ago +1 -0 lines changed
grok4_system_turn_prompt_v8.j2