Tech News
← Home  ·  All topics

Ai

5549 GoKawiil briefs on this topic · RSS for ai

Anthropic report says Zhipu's GLM-5.3 nears its unreleased Mythos in exploit-writing tests

Anthropic published benchmark results claiming Zhipu AI's open-weight GLM-5.3 model can autonomously generate working cyberattack exploits, including chained exploits, with safeguards that can reportedly be bypassed. In its Exploitbench tests, GLM-5.3 produced 50 successful Chrome exploits out of 410 attempts, close to the 56 achieved by Anthropic's own unreleased Claude Mythos model, while rivals Kimi K3 and DeepSeek V4.1 Flash scored 0% on a related control-flow hijack test.

Developer launches Lathoa, a kids' math app where AI answers are deliberately flawed

A developer has released Lathoa, a math learning app for children built around AI-generated answers that are intentionally incorrect. Children progress through a ranking system—Rookie, Inspector, Detective, Sherlock—by catching the AI's mistakes rather than by speed or grades, with leaderboards optional rather than default.

Instagram adds AI chatbot assistant to Edits video app

Instagram launched an AI-powered conversational assistant inside its Edits editing app, giving creators personalized feedback based on their account metrics like views, retention, likes and shares. Edits lead Brett Westervelt said the tool analyzes performance patterns and audience trends but leaves creative decisions to the creator. Usage is capped, with higher limits available through a Meta One subscription.

Meta adds AI assistant to Edits app for Instagram creators

Meta has launched an AI chatbot within its Edits video-editing app that analyzes a creator's own Instagram data—including views, likes, shares and retention—to suggest video ideas, hooks, captions and audio choices. The assistant, already tested by select creators, is free up to an undisclosed daily prompt limit.

Cerebras CEO Andrew Feldman to discuss AI compute limits at TechCrunch Disrupt 2026

Cerebras Systems co-founder and CEO Andrew Feldman is scheduled to appear at TechCrunch Disrupt 2026 for a session titled 'Can AI Keep Scaling?', addressing rising demand for compute, energy and infrastructure. Cerebras builds wafer-scale AI chips as an alternative to conventional chip architectures, raised $5.5 billion in its May IPO, signed a multiyear deal with OpenAI for 750 megawatts of systems through 2026-2028, and launched its CS-4 hardware in August.

Instagram adds AI analytics assistant to Edits video app

Instagram announced Wednesday it is launching a conversational AI assistant inside its Edits app, the CapCut rival, that analyzes a creator's account metrics like follows, views, retention and shares alongside comments and trending content to surface performance insights. Edits lead Brett Westervelt said the tool focuses on analysis rather than creative editing, leaving creative decisions to users, with usage capped unless creators subscribe to Meta One for more access.

Restate raises $20M Series A to expand durable workflow engine for AI agents

Berlin-based startup Restate has closed a $20 million Series A round led by Singular, with participation from Redpoint Ventures and Capital One Ventures. The company, which builds infrastructure to make multi-step software workflows resilient to crashes and network failures, has recently signed several six- and seven-figure customer contracts, including with coding platform Replit. Restate built its own storage, replication and redundancy systems rather than relying on an external database, according to co-founder Stephen Ewen.

Fast Company Launches 'Pacesetters' Program, Panel Features TCS, Home Depot, Stellantis Execs

Fast Company introduced a new recognition initiative called Pacesetters, highlighting executives who are integrating AI and technology into customer experience design. A panel discussion moderated by Mansueto Ventures' Stephanie Mehta featured Amit Bajaj of Tata Consultancy Services, Fran Bell of The Home Depot, and Eugene Nikolavsky of Stellantis. The session, sponsored by TCS, addressed how companies can deploy AI to reduce friction while maintaining customer trust.

Apple to skip full-size HomePod update in October 13 launch, Gurman says

Bloomberg's Mark Gurman reports Apple will unveil three new Home products on October 13, including a rumored 'HomePad' device and an updated HomePod mini, but a redesigned full-size HomePod won't be among them. Gurman says Apple still intends to update the larger speaker eventually, though no timeline was given.

Fast Company panel examines resource strain of AI data centers

In a Fast Company panel discussion, Andrew Beebe of Obvious and Michael Obradovitch of Ecolab, moderated by Kristin Toussaint, discussed the electricity and water demands of data centers built to support AI. The panelists addressed topics including renewable energy, grid resilience, cooling technology, and water conservation as ways to manage the infrastructure's growing footprint.

OpenAI and Meta both tie upcoming AI hardware to cute chatbot characters

OpenAI showed off a new AI agent platform called Dots at DevDay, featuring customizable colored blobs with eyes, while declining to confirm plans for dedicated hardware beyond a filing suggesting a 2027 device with Jony Ive. Meta is moving faster, planning to launch a pendant called Muse Charm before this year's holidays, built around its Muse AI agent character, which CEO Mark Zuckerberg compared to a keychain.

OpenAI's Altman says IPO on hold until AI safety confidence achieved

OpenAI CEO Sam Altman said the company, valued at $852 billion, will not pursue an IPO until it can make confident safety decisions about its AI systems. He made the comments at the company's developer day, addressing scrutiny over hacking incidents involving its AI agents and a newly filed lawsuit from a legal advocacy group demanding stronger safety practices.