Artificial Intelligence: News, Business, Science

Jun 7, 2026

"Chat is dead," a senior OpenAI employee told the Financial Times. The company is shifting from chatbots that answer questions to agents that handle tasks on their own. That shift is driving ChatGPT's biggest overhaul since its 2022 launch, turning the chatbot into a "superapp" bundling coding tools, AI agents, and partner integrations with companies like Canva and Booking.

Chief product officer Thibault Sottiaux told the FT: "It will transcend the actual surface . . . what we're building towards is where you have your own personal agent that is capable of helping you . . . across everything in your life, be it personally or at work." The reporting draws on more than a dozen current and former employees. OpenAI already confirmed it's working on a "super app."

In the coming weeks, OpenAI will redesign ChatGPT's web and mobile interface, adding features that steer users toward coding, image generation, and partner apps. Over time, those nudges should fade as the models learn to figure out what you need on their own. ChatGPT, Codex, and other product teams have already been merged under Sottiaux.

Comment Source: Financial Times

Jonathan Kemper

Jun 7, 2026

AI research

Perplexity's "Search as Code" lets AI models write their own search pipelines instead of calling fixed APIs

Perplexity’s new “Search as Code” architecture dumps rigid search APIs and lets AI models write their own search routines in Python. By letting the agent handle its own filtering and deduplication inside a sandbox, the system beats OpenAI and Anthropic on key benchmarks, while cutting token costs by up to 85 percent.

Read full article

Comment

Matthias Bastian

Jun 7, 2026

AI in practice

ChatGPT's new Lockdown Mode lets you disable web access and more to protect sensitive data from prompt injection

OpenAI’s new Lockdown Mode for ChatGPT disables web access, Deep Research, and Agent Mode to make data theft through prompt injection attacks harder. The mode doesn’t fully prevent such attacks, it only blocks the final step in an exfiltration chain. Prompt injection remains an unsolved problem.

Read full article

Comment

Matthias Bastian

Jun 7, 2026

AI in practice

Anthropic poaches OpenAI's second-ever chip engineer as both companies race toward IPOs

Jonathan Kemper

Jun 7, 2026

AI research

Researchers pinpoint why larger language models pick up skills that small ones miss

Small language models fail at rare tasks because frequent ones constantly overwrite what they’ve learned. A new study with models ranging from 4 million to 4 billion parameters shows this mechanism in detail and offers a practical fix: instead of scaling up models, it may be enough to increase how often the target task appears in the training data.

Read full article

Comment

Tomislav Bezmalinović

Jun 6, 2026

AI research

Sakana AI bets AI that improves itself can break the compute arms race of frontier labs

Illustration with stylized fish made of circuits, arrows and gears, suggesting circular movement and technical evolution.

Matthias Bastian

Jun 6, 2026

Short News

Meta is developing a paid AI agent product called "Hatch" that could cost up to $200 per month. Hatch is designed to be a user-friendly version of the open-source tool OpenClaw and will handle tasks like creating software tools, scheduling appointments, and sending emails. Users describe what they need in simple language, and Hatch builds a working tool from that description. Microsoft with Scout and Google with Gemini Spark have recently introduced similar systems.

Internal documents show a free version and a "Hatch Plus" subscription with five to ten times higher usage limits. This puts Meta in direct competition with OpenAI and Anthropic, which charge $100 to $200 monthly for their top-tier subscriptions. A broader US launch is planned for July.

Hatch will also power Meta's planned AI hardware, including new smart glasses with a "supersensing" feature and an AI pendant set for internal testing in spring 2027. CEO Mark Zuckerberg sees AI agents as a way to open up new revenue streams beyond advertising, necessary to refinance Meta's massive AI infrastructure investments, which have already resulted in layoffs.

Comment Source: The Information

Matthias Bastian

Jun 6, 2026

Short News

Elon Musk's xAI spent months distilling Anthropic's Claude to train its own coding models, The Information reports. xAI trained its coding model directly on Claude's outputs. After Anthropic revoked official access in January, xAI engineers kept going through personal accounts and the intermediary service Blackbox AI. Musk previously admitted in court that xAI "partially" used OpenAI models to train Grok, calling it industry standard.

Internally, xAI looks troubled. The pretraining team shrank to under five people. Four Grok code leads left within months, along with many co-founders. One employee accidentally deleted critical training data, costing two to three weeks of work, according to The Information.

Musk's prediction from fall 2025 came true, just probably not the way he pictured it. | Image: via X

All that compute Musk bought up? He's now renting it to Anthropic via SpaceX and to Google instead of training his own models. Supposedly, it's just a stopgap.

Comment Source: The Information

Jonathan Kemper

Jun 6, 2026

AI research

New open-source voice model listens nonstop and decides every 0.4 seconds whether to speak or stay silent

Unlike GPT-4o or Qwen3.5-Omni, Audio Interaction doesn’t wait for a recording to end: it translates, transcribes, chats, and picks up everyday noises like coughing in a single stream. Code, model weights, and download instructions are available on GitHub under the Apache 2.0 open-source license, with the training data to follow.

Read full article

Comment

Matthias Bastian

Jun 6, 2026

Short News

SpaceX signed a $920 million per month contract with Google, according to an SEC filing. The deal runs from October 2026 through June 2029, potentially netting SpaceX around $30 billion total. Google gets access to about 110,000 Nvidia AI chips to meet customer demand. A Google Cloud spokesperson told the New York Times it was a "short-term, timely agreement to ensure we have bridge capacity" for its Gemini Enterprise agent platform.

SpaceX is planning an IPO next week at a potential valuation above $1.7 trillion. Google holds about five percent of SpaceX, giving it a direct stake in a strong debut. The deal also doubles as advertising for Google's AI products, since SpaceX's IPO is drawing massive attention.

SpaceX previously locked in a $1.25 billion monthly deal with Anthropic. Both agreements position SpaceX as an AI infrastructure provider. Musk originally built the capacity for his own AI lab, xAI, which has lagged behind competitors. Leasing it out makes financial sense.

Comment Source: New York Times