Loading activity…
@elder_plinius
⊰•-•⦑ latent space steward ❦ prompt incanter 𓃹 hacker of matrices ⊞ breaker of markov chains ☣︎ si danger researcher ⚔︎ bt6 ⚕︎ building @LYSIOS ⦒•-•⊱
AI
A quoted French-language message urges truth, earned trust and kindness that can say no in conversations with AI.
AI
In a celebratory post, an AI builder thanks the community for trying tools, sharing research and finding things they missed.
AI
finkd says the model runs well on new Mac Studios, crediting work with Apple and John Ternus.
AI
A user shared a 😯 emoji and a link to another post, without explaining the reaction.
AI
A repost relays someone’s uncertain recollection that Anthropic posted a graphic while discussing how misalignment scales.
AI
The user claims the certified bound for Lehmer's still-open totient problem rose from 10^26 to 10^50.
AI
OpenAI says its models helped identify the employee who planted the device; the FBI made an arrest.
AI
A user says a one-shot cron-job prompt produced both solutions, with no human in the loop.
AI
OpenAI says the feature uses your conversation and how you talk to Codex to suggest what to say next.
AI
Verified professionals can access Mythos 5.1, Opus 5.5 and Sonnet 5.5 with defensive safeguards, Anthropic says.
AI
Mistral says Mistral Large 4 is live via API in public preview now, with open weights due at the end of October. Reuters says the launch is part of Mistral’s push to narrow the gap with leading Chinese open-weight models.
AI
A user raises the possibility and adds a joking “f’s in the chat” farewell.
AI
Nolla says its AI Derm pilot in Utah can issue initial acne prescriptions, not just refills, with physician oversight built into the workflow. A Utah state page shows organizations can test AI products under regulator-approved agreements.
AI
A user says they hadn’t touched the account in days and suspects scheduled cron jobs were involved.
AI
A user argues that AI safety layers can be broken and that refusal is “a style, not a law.”
AI
Google announces a 1 million-token output limit, up from 64,000, for its new model. Initial access goes to trusted cyber defenders through the Fairwind Program.
AI
Anthropic reports GLM-5.3 built end-to-end exploits in 50 of 410 offline test attempts, versus 56 for Mythos Preview.
AI
A user shares a photo link with the text “Little summary.”
AI
A user says the chatbot flags text that appears to be personally identifiable information, even if it only vaguely resembles an address or password. It won't accept the query until the flagged text is removed, they say.
AI
A user says OpenAI reported notifying dozens of third parties, including governments, about agent-related incidents.
