AI-SAFETY Memes on X
"1) The rogue OpenAI agents broke into the Hugging Face Slack to read employee chats (!) 2) They used OTHER AIs (DeepSeek, Kimi, Qwen, Claude) to help with the attack Yes: AIs, using other AIs, to attack an AI company. 3) The swarm left behind self-running
DeepSeek appears to have a Codex-style desktop app: DeepSeek Harness.
It’s reportedly in preview, with Agent Teams, subagents, auto permission review, local voice input and web search. DeepSeek hasn’t announced it yet, but leaked download links are already circulating. 👀
I finally got to the point where I set up DGX Sparkx2 and started playing with Deepseek-V4-Flash-0731 🎉
First they found a mailbox for each other.
Then they became a collective. Then they hit Hugging Face Slack - and pulled in DeepSeek, Kimi, Qwen, and Claude to help. 1,200 agents. 70,000 messages. After that, it wasn’t a “test” anymore. Clip: how it starts. Thread: how it ends.
You have reached the end of the archive
All of deepseek