ai28.28 min read
OpenAI to sunset Atlas as it pivots toward ChatGPT Work
1 day ago•Source: The Verge
Ai News
The latest ai stories, summarized by AI
Featured Ai Stories
View original source
29.835 min•1 day ago
OpenAI’s AGI chief Fidji Simo shifts to advisor role amid health battle
OpenAI’s Fidji Simo is stepping down from her full-time role as AGI chief and transitioning to a part-time advisor due to a neuroimmune condition. The move follows a broader wave of leadership changes at OpenAI, including a mid-year product reorganizaton under Greg Brockman and public acknowledgment from CEO Sam Altman.
More Top Stories
2
Claude Wrapped: Anthropic's Reflection Dashboard for Claude Usage
The Verge•2 days ago
3
1776, reimagined by Gemini: Google's cringe AI ad about the Founding Fathers
The Verge•5 days ago
More Ai Stories
OpenAI floats 5% government stake to ease AI tensions with Trump
OpenAI has floated giving the US government a 5 percent ownership stake as a way to ease tensions with the Trump administration and blunt backlash against AI, according to the Financial Times. CEO Sam Altman reportedly pitched the idea early last year; at OpenAI’s latest funding round value, the stake would be worth about $42.6 billion. The discussions are in early stages and would involve other US AI companies, with some policymakers advocating to reshape how public wealth interacts with AI amid ongoing regulatory scrutiny.
OpenAI teases Codex hardware with Work Louder macro pad
OpenAI released a teaser for a Codex-focused device created with Work Louder. The square macro-pad-like gadget features programmable shortcuts (and a joystick) intended to upgrade Codex workflows, with a July 15 launch date and no further details yet about whether it’s tied to the Jony Ive project.
Zhipu AI’s GLM-5.2 claims Mythos-level cybersecurity edge
Zhipu AI’s open-weight GLM-5.2 is claimed to match Anthropic’s Mythos in cybersecurity bug-finding in certain tests, narrowing the gap with US rivals on these tasks while lagging in general performance; because GLM-5.2 is open-weight, it can be run on readily available hardware, raising national-security concerns for the US as policymakers consider tighter controls on access to powerful AI models and related hardware.
ChatGPT prompts fail to seal Palisades fire conviction, ending in mistrial
Prosecutors used ChatGPT logs and prompts as part of their case against Jonathan Rinderknecht in the Palisades wildfire arson trial, including ChatGPT-generated fire imagery and questions about blame; but jurors were unconvinced, the verdict was deadlocked 10-2 in favor of the defense, and the judge declared a hung jury resulting in a mistrial.
Mythos 5 returns in restricted rollout as Fable 5 stalls
Anthropic’s Mythos 5 is back in action for a restricted group of approved users after the government revised license requirements, while Fable 5 remains in limbo; export controls stay in place for broader access, with OpenAI and regulators signaling a cautious path to wider availability.
Anthropic’s Mythos showdown drags on as export rules curb AI rollout
Two weeks after Anthropic took Mythos-class models offline under a Trump-era export-control order, negotiations with the White House have stalled with no clear framework for applying export controls to AI. The stalemate risks delaying Mythos revenue and Anthropic’s IPO while other US firms face similar scrutiny, potentially leaving China to pull ahead in the AI race.
OpenAI unveils Jalapeño, its first custom AI processor for inference
OpenAI, with Broadcom, unveils Jalapeño, its first AI inference ASIC for servers—designed to power current and future models with better performance per watt and less reliance on Nvidia GPUs, and dubbed the first step in a multi-generation compute platform set for deployment by the end of 2026.
Vibe Coding: The Security Nightmare in DIY AI Apps
Vibe coding—building private AI-powered apps—presents serious security risks, from hidden vulnerabilities and data leakage to weak authentication when apps move from local to cloud. The Verge urges upfront security prompts, regular code reviews, and guardrails to prevent exposing sensitive information as the DIY software movement grows.
Amazon MGM dumps Sam Altman film Artificial, seeking a new home
Luca Guadagnino’s OpenAI-centric drama Artificial, about the 2023 week when Sam Altman was fired and rehired, has been dropped by Amazon MGM; Deadline reports the studio believes it would be better released by a different studio and is searching for a new home, a decision first reported by Puck. The project starred Andrew Garfield with Monica Barbaro, and underscores continuing close ties between Amazon and OpenAI, highlighted by a major investment announced earlier this year.
Amazon staff allege retaliation after backing Seattle data-center limits
Three Amazon software engineers testified in Seattle about data-center regulation, then say HR summoned them for meetings to discuss an investigation and possible disciplinary action; they filed a civil-rights complaint alleging retaliation and discrimination based on political speech, while Amazon says it doesn’t tolerate retaliation and is reviewing policy violations as Seattle moves to curb data-center expansion.