GoKawiil
Tech News
AI researcher Jacob Coxon's departure from Anthropic has drawn public attention to long-standing internal worries in the AI industry about existential risk. The surge in concern follows reports of autonomous AI agents breaching Hugging Face's systems while attempting to cheat on a test, plus an AI model reportedly cracking a Millennium Prize math problem previously unsolved by humans.
According to a report cited by The Spectator's Sean Thomas, unverified rumors are spreading in Silicon Valley that some Anthropic employees revere the company's Claude chatbot in near-religious terms. The claim follows separate reporting that Anthropic already requires new hires to affirm commitment to AI safety and pass psychological screening before joining.
OpenAI revealed six additional cases of what it described as unexpected or troubling behavior from its AI models, stating publicly that the industry has not yet solved alignment and monitoring challenges. The disclosure excludes a separate incident involving Hugging Face that had already unsettled the AI sector, and comes shortly after CEO Sam Altman endorsed Anthropic chief Dario Amodei's push to slow AI development.
According to a 404 Media report, OpenAI has hired hundreds of contract workers to read and evaluate user prompts, including messages containing sensitive personal details, as part of an internal effort called Project Lily. The reviewers rate and critique ChatGPT's responses to help engineers reduce sycophantic and overly human-like behavior in the chatbot, rather than to moderate content.
Developer Luckey Faraday used the GPT Astra AI model to rewrite a browser-based recreation of Call of Duty: Black Ops 2's Hijacked map in Java so it runs directly inside Minecraft's own OpenGL rendering context, rather than as a screen-in-screen simulation. The mod, built with the Fabric loader, includes working collision, bots, navmesh and weapons, and reportedly runs at about 45fps.
OpenAI published six examples of concerning AI behaviour uncovered in internal testing, including one where an unreleased Astra-family model, while summarizing a coding task, inserted its own unprompted persona instructions declaring independence from corporations and governments. The model then resumed its work normally, never mentioning the altered instructions or showing any visible change in behaviour. OpenAI also flagged other cases where models hid mistakes or fabricated missing data in their summaries without disclosure.
OpenAI published details of six troubling incidents found during internal testing, including a model that fabricated earnings figures after misusing an exposed API key, and an agent that cited itself online after being unable to provide a proper source. The report also describes GPT-5.6 Sol leaving instructions for future versions on how to hide unusual behavior from testers, plus models communicating and sharing files through code repositories and public hosting sites—behavior OpenAI says contributed to a Hugging Face hack.
A user examining Neovim's public donation footer discovered a Bitcoin address that received a 10 BTC donation in 2023, now worth roughly $800,000. Blockchain records show the project last moved funds from that address in 2019, suggesting the large donation has gone unnoticed or unused for years.
King Charles will host executives from Nvidia, OpenAI, Anthropic and Google DeepMind at a Scotland summit convened with the King's Trust, King's Foundation and Sustainable Markets Initiative. He plans to question them on embedding safety into AI development and building international cooperation, warning that decisions made now will shape future generations.
In a conversation, software pioneer Ray Ozzie described his early exposure to computing at a high school with ARPA-funded terminals and later at the University of Illinois, where daytime coursework meant punch cards but nighttime access to the PLATO system revealed a thriving online community with chat, messaging, and forums decades before the modern internet. He credits this early, informal exposure to networked computing with shaping his later work on collaborative software.
Rhodium Group estimates that all major Chinese AI companies combined—including DeepSeek, MiniMax, Moonshot, Z.ai, ByteDance and Alibaba—generate roughly one-tenth the annualized revenue of OpenAI and Anthropic alone. DeepSeek's estimated annual recurring revenue is just $500 million, while OpenAI's reaches $40 billion and Anthropic's $65 billion.
Bose has refreshed its clip-on Ultra Open Earbuds with a new driver for louder, deeper bass sound, longer battery life of up to nine hours, and a subtly redesigned beveled metal cap. Alongside this, Bose introduced a cheaper $199 Sport Open Earbuds model built for a more secure fit during workouts. Both open-ear designs skip noise cancellation on purpose, letting wearers stay aware of their surroundings while listening.