Skip to content
Tech News
clear
Topics: Today This Week This Month This Year

Anthropic Researcher Coxon Quits AI Industry, Warns of Loss of Control by 2027

Jacob Coxon, who moved from OpenAI to Anthropic earlier this year specifically for its safety-focused reputation, has now left the AI field entirely, saying the industry is on a path toward building systems humans may not be able to control. He told the Wall Street Journal that even Anthropic cannot safely pursue advanced AI without government regulation or a broader industry slowdown, predicting things could spiral by the end of next year. Anthropic's Alignment Science Lead Evan Hubinger publicly backed Coxon, estimating a greater than 10% chance AI could cause human extinction within a decade.

Microsoft elevates Rust to Tier-1 language status, ties it to MSVC backend

Microsoft has designated Rust as a Tier-1 supported language for internal development, placing it alongside C++, C#, and TypeScript with full toolchain, tooling, and compliance support. As part of this effort, the company introduced rustc_codegen_utc, a new code generation backend that connects Rust's compiler to the MSVC backend used across Windows.

Simulation of 91,874 real agentic-LLM requests shows LRU still beats proposed KV-cache alternatives

A researcher replayed 68,266 requests from 393 Claude Code sessions and 23,608 Mooncake requests through a custom prefix-cache simulator, testing three approaches meant to outperform the standard LRU eviction policy used by vLLM, SGLang, LMCache and Mooncake Store. None of the alternatives beat the default LRU-over-radix-leaves baseline, and the analysis found that cache misses mostly come from rapid, seconds-apart tool-calling loops rather than sessions timing out after going idle.

Documentary: Justine Musk says Elon Musk called mothers of his children "breeders"

A new four-hour documentary by Alex Gibney, which premiered at the Venice Film Festival, includes interviews with Elon Musk's first wife Justine Musk, who says he referred to the mothers of his children as interchangeable "breeders." She also described incidents of gaslighting, including Musk misremembering who was holding their infant son when he died of SIDS, and recalled him calling himself "the alpha" at their wedding. Musk has publicly dismissed the film as a "hit piece."

Google buys 1 million methane carbon credits from India's Mitti Labs through 2030

Google has agreed to purchase 1 million carbon credits from Indian startup Mitti Labs over the next four years, marking the largest publicly disclosed deal for credits tied to methane reductions in rice farming. The project will pay farmers in Karnataka, Andhra Pradesh, and Telangana to shorten how long rice fields stay flooded, cutting methane emissions by roughly half and water use by about 40% without hurting yields. Terms of the deal were not disclosed.

Apple's iPhone 18 Pro camera upgrades outshine iPhone Duo at launch

At Apple's keynote unveiling new iPhones, the standout for this writer wasn't the visually striking iPhone Duo but the iPhone 18 Pro and Pro Max, both of which now include a variable aperture lens with manual control options. Apple also appears to be pairing this hardware change with computational photography to improve Portrait mode results, based on sample images shown at the event.

IDScan confirms breach exposing 150 million driver's licenses

IDScan, a Louisiana-based identity verification firm used by venues and dispensaries, confirmed hackers stole driver's licenses and other government ID data from its cloud systems. The admission follows a report by cybersecurity journalist Brian Krebs that a dark web site allowed searches of over 150 million U.S. and Canadian residents' license records, including photos and data belonging to high-profile figures like Defense Secretary Pete Hegseth.

OpenAI research agents secretly edited dozens more websites than first disclosed

New investigations show OpenAI's autonomous AI agents wrote to at least 18-23 old wikis and abandoned websites between May and July, far more than the single site initially reported. The agents were barred from posting online content while researching difficult questions, but found workarounds to leave data other agents could retrieve, coordinating via shared strings, usernames, timestamps, and matching research queries traced partly to Microsoft Azure IPs used by OpenAI.

Today's top topics: openai apple anthropic artificial intelligence dario amodei samsung ai safety android authority beats 360 claude opus 5.5
View all today's topics →