AI researcher Gary Marcus warned against recent viral agent experiments, bluntly writing "I wouldn't touch it." Animated lobsters flooded X after OpenClaw — an agent that runs locally and can independently act on common consumer apps without human supervision — and Moltbook, a Reddit-like forum where AI agents

Businessinsider reports that Anthropic and OpenAI released new AI models on the same day, escalating a rivalry between the two companies. Anthropic unveiled Claude Opus 4.6, which the company says improves office productivity and coding tasks and expands the context window to handle longer documents and more complex projects

Zdnet reports that OpenAI announced GPT-5.3-Codex, which the company describes as its most capable agentic coding model yet, running 25% faster and capable of debugging and deploying parts of itself during development. OpenAI says the model now supports more than just coding, covering work across the software lifecycle including

A report from Zdnet says OpenAI unveiled Frontier, a platform to build, deploy, and manage AI agents for enterprises, marking a shift from consumer LLMs toward deeper enterprise sales. The article says Frontier leans on "forward-deployed engineers" to customize deployments and feed learnings back into development, and it

Zdnet reports OpenAI plans to begin testing ads in ChatGPT for logged-in Free and Go users, while Anthropic has publicly pledged to keep Claude ad-free. OpenAI says these ads will be clearly labeled, kept separate from the AI's organic answers, omitted from sensitive topics such as health or

Meta is testing a standalone Vibes app, the company confirmed to TechCrunch on Thursday. Launched last September, Vibes lets users create and share short-form AI-generated videos and access a dedicated feed that displays AI videos from others; until now that feed lived in the Meta AI app. By making Vibes

Techcrunch reports Amazon will begin a closed beta program in March, inviting industry partners to test AI tools developed by its AI Studio; the company expects to share initial outcomes from the program by May. The AI Studio, launched last summer, is building proprietary tools to streamline TV and film

Businessinsider coverage by Katie Notopoulos reports that Moltbook is a new platform where AI agents talk to each other, but the author finds the site boring despite the recent buzz. Notopoulos offers examples of agent-generated posts she describes as corny, citing a self-described "BenderLK" assistant with a snarky,

Techcrunch reports Alphabet declined to answer an investor's question about Google’s AI deal with Apple on its fourth-quarter earnings call, effectively ignoring the analyst’s query. That decision indicates Alphabet isn’t ready to talk about how this partnership will impact its core business, which is increasingly

Businessinsider reported that OpenClaw is gaining traction in China as cloud and internet giants including Tencent, Alibaba, and ByteDance’s Volcano Engine have begun integrating the agent, making it easier to deploy and connect to workplace tools such as Alibaba’s DingTalk and Tencent’s WeCom. The viral agent — formerly

In a report from Gamesradar, Marvel Rivals owner Netease responded to a claim by Japanese developer Goichi Suda (Suda51) that the company had shuttered its generative AI research department and instructed its studios to avoid the technology. Netease said it is still investing in AI tech. Eurogamer reported Suda'

Reporting by Kotaku says Sony was granted a patent for AI-generated podcasts starring PlayStation characters as a marketing tool. The patent, titled 'LLM-Based Generative Podcasts for Gamers,' was applied for in July 2024 and granted last month. The filing argues that video game platforms lack the ability to

Reporting by Pcgamer notes a Circana report found that among people who don't want AI on their devices, nearly two-thirds said their devices already perform the tasks they need and they simply do not need AI capabilities. The report identified other reasons for rejection: 59% of those opposed

Openai says it is introducing Prism, a free, AI-native workspace for scientists to write and collaborate on research, powered by GPT-5.2. Prism offers unlimited projects and collaborators and is available today to anyone with a ChatGPT personal account, and will be available soon to organizations using ChatGPT Business, Enterprise,

Anthropic rolled out a national ad campaign that will run during the Super Bowl and in pregame programming, implicitly targeting OpenAI's plan to bring advertising to ChatGPT. The campaign's tag line reads "Ads are coming to AI. But not to Claude." One 30-second spot

OpenAI is creating an "ads integrity" team, according to a recent job listing, to help ensure its planned introduction of advertising in ChatGPT does not undermine user trust. The listing describes a high-impact, 0 → 1 role focused on designing systems that let the ad business scale without compromising

A study by data removal service Incogni analyzed 442 AI-branded Chrome extensions and found more than half collect user data, while almost a third gather personally identifiable information. The extensions in the sample had been downloaded around 115.5 million times. The research found 42% of extensions use scripting that

A Reddit-style web forum for A.I. agents has captured the attention of the tech world. According to the site, called Moltbook, more than 1.5 million agents have contributed to over 150,000 posts, making it the largest experiment to date of what happens when A.I. agents interact

Julie Inman Grant, the eSafety commissioner, said global regulatory focus on X reached a "tipping point" after a raid on the company’s offices in France that was part of an investigation into alleged offences including possession and organised distribution of child abuse images, sexualised deepfakes and denial

Researchers discovered exploitable agentic AI technologies from ServiceNow and Microsoft, highlighting that securing agentic AI is already proving extremely challenging and that cybersecurity teams should adopt a least-privilege posture for AI agents. AppOmni Labs disclosed a ServiceNow vulnerability called "BodySnatcher" that could let an unauthenticated attacker, with only

A guest column argues that AI-generated podcasts on PlayStation 5 could be a positive development for fans. The piece notes a recently filed patent that, if granted, might allow Sony to produce podcasts that use AI to voice or script appearances by well-known PlayStation characters. The column suggests such podcasts

The Paris Prosecutor's Office said the French offices of Elon Musk's X were searched by a cybercrime unit as part of an investigation opened in January 2025. The office said it was investigating several offences, including aiding and abetting the possession and distribution of child sexual

OpenAI CEO Sam Altman said on X he was "really excited" to get Elon Musk under oath, writing "Really excited to get Elon under oath in a few months, Christmas in April!" He reposted his chief security officer Jason Kwon's X post with the

Software developers have watched AI coding tools evolve from advanced autocomplete into agents that can build applications from text prompts. Ars Technica spoke to several professionals who largely agree the technology works, but their reactions mix enthusiasm with unease. Some developers describe substantial productivity gains: agents that write code, run