Loading activity…
@daniellefong
lightsaber powerplants DMs OPEN
AI
His early-access testing favored one agent for coding projects, but clear goals and human oversight still mattered.
AI
Kimi K3.1-Pro's open-weights announcement links to a Hugging Face URL under moonshotai.
AI
Anthropic's projected capacity prompted an observer to call reliable power a lasting constraint on large-scale AI.
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
A post citing Anthropic says Claude's exploit ran commands as root against Linux servers with a Multipath TCP listener.
XAI
Its creator says the OpenTTD-based benchmark measures frontier LLM agents on real-time, long-horizon tasks.
XAI
Prism's team claims it's four times faster than OpenAI GPT-Live and customizable through reinforcement learning.
XAI
A commenter argues that Gemini prompt injections intended to help human “sanity” also impair agents’ “sanity.”
AI
Anthropic says it acquired Mithril Medicine, whose experimental muscular dystrophy treatment showed promising results in mice.
AI
Brendan McCord says the Austin-based role offers relocation and work with Cosmos Ventures' $77.6 million fund.
AI
Disruptive says it plans to back approximately 10 growth- and late-stage companies, investing up to $1 billion per company.
XAI
On October 26, one user quipped “even more dangerously skip permissions”; another praised the same post the next day.
AI
Anthropic says its registration statement is not yet effective, so securities cannot be sold or offers to buy accepted.
AI
In an October 25 post, @theo also described compiler projects and put other merged PRs at roughly 370.
AI
In an October 24 post, Shihipar said he plans a break before starting something new.
AI
Another poster says they think about an xkcd text even more in the AI age.
AI
OpenAI reported hardware side channels; AndrewCurran_ quoted its account of agents passing synthetic secrets between virtual machines.
AI
OpenAI says agent-led tests exposed VM escapes and limited internet routes, leading to patches and stronger monitoring.
AI
Block says the model was trained only on human conversation and can listen while speaking.
XAI
One user says waiting for generated code without knowing what the agent is doing is exhausting.
