All Stories — Page 98

Haritz Puerto says a paper on decomposing and measuring evaluation awareness just dropped, plus a resource called EvalAwa..., but links and authorship were not shared.

In a new paper and open releases, Li and collaborators decompose evaluation awareness, test nine models across four benchmarks, and publish a factor-controlled dataset and code.

The tally counts only Hugging Face and Ollama pulls, hinting at on-device demand but leaving methodology and release timing unclear.

XRoboHub reports the line is live at an integrated Shenzhen facility, but capacity timing and T800 specs were not disclosed.

Coffee meetups land June 2nd, 4, and 6, with RSVPs running through Luma as Asimov convenes folks interested in humanoids and AI.

The company targets high-performing professionals, pairing FDA-cleared TMS with d-cycloserine and optional $200 per month maintenance sessions.

Ex-ESA data science intern and Volvo Group engineer turn to agent infrastructure; Box targets agent developers with one-command sandboxes, a simple API, and pricing claims 5x below popular offers at equal specs.

Demos show a Mac agent that records your cross-app workflow once, compiles it into a deterministic DSL, and replays it locally with zero recurring tokens. With freu-cli open-sourced and a local vision execution model coming, can ahead-of-time semantic compilation beat screenshot agents on cost and latency?

Socket researchers say the TrapDoor campaign planted credential-stealing payloads in more than 34 packages and 384+ versions, targeting crypto and AI developers.

Google DeepMind reports a full-featured AlphaProof Nexus agent solved 9 of 353 open Erdos problems at a few hundred dollars per problem and proved 44 of 492 OEIS conjectures; code and Lean proofs are on GitHub. This is an arXiv preprint and community validation is pending.

Zhou's team links raw EHR, web, and chest X-ray tools, and reports open-source SOTA after distilling Claude Opus 4.6 into a 35B model.

In a rare interview on The Knowledge Project, the OpenAI co-founder recounts quitting within hours, sketching a backup company, and why they stopped showing reasoning traces.

The app says it now runs a lightweight local LLM as an interrogator model and splits generation and editing into distinct modes for a cleaner workflow.

Aligned News flagged the move; Wikipedia lists Karpathy on Anthropic’s pretraining team in 2026, underscoring top-tier talent consolidation at frontier labs.

Founder-CEO Caroline Walerud is stepping up field validation and adding 12 technician and engineering roles as Northzone leads a convertible from existing backers.

The San Francisco afterparty will spotlight 100+ expert curated agent tasks, live demos of the BenchFlow SDK, and Kaggle’s new Agent Benchmarks.

The longtime founder-operator says coverage has drifted from builders and pledges to chase early product signals over polished funding cycles.

Cloudflare alone found 2,000 issues; Anthropic’s OSS scans show a 90.6% triaged true-positive rate and project nearly 3,900 high-or-critical bugs ahead.

Michael Grinich unveiled auth.md with early support from Cloudflare and Firecrawl, giving apps a standard way to onboard AI agents without brittle browser hacks.

Less than two years in, the industrial robotics startup led by Apple alum Hamza Derbas is rumored to be pursuing a $100M Series A.

In a 10-iteration code-and-rewrite challenge, the team let each model build and train a Tetris bot, then compared the final agents.

In a thread on X, Nav Toor said Frigate, an MIT-licensed NVR app, offers no cloud, no subscription, and privacy by default, citing 32,057 GitHub stars.

Approaching $10M ARR, Cera says Polsia runs companies autonomously with one founder and AI, and even handled its own raise while he "just showed up for signatures."

The public GitHub project under superset-sh is shipping agent skills, a Linear ticket draft flow, and editor UX tweaks, with 10.9k stars and nearly 2,900 commits.