OpenAI on IT News

OpenAI's Chief Futurist Joshua Achiam Departs After Nine Years

Wed, 08 Jul 2026 00:00:00 +0800

Joshua Achiam, OpenAI’s chief futurist and one of its longest-serving safety researchers, announced Tuesday that he will leave at the end of July — ending a nearly nine-year run at the AI company.

Achiam joined OpenAI in 2017, back when the organization was still operating as a nonprofit research lab. He went on to co-author the GPT-4 technical report and became a core voice inside the company on AI safety, long-term risk, and public policy. His departure is not tied to any single event, he said, but was a decision he arrived at after long reflection.

OpenAI's UK "Stargate" project was a PR stunt, investigation reveals

Sun, 05 Jul 2026 00:00:00 +0800

There’s a familiar rhythm to big AI infrastructure announcements — the press conference, the handshake photo, the round number in billions. But sometimes the script falls apart. The UK’s so-called Stargate project, touted as OpenAI’s flagship British data center and a cornerstone of US-UK tech cooperation, was shelved in April. OpenAI blamed high energy costs and regulatory uncertainty. The real story, uncovered by The Guardian, is messier.

Meta's 'Watermelon' AI Model Has Caught Up to GPT-5.5, Exec Says

Fri, 03 Jul 2026 00:00:00 +0800

Meta’s next-generation AI model, codenamed Watermelon, has matched OpenAI’s GPT-5.5 in key benchmarks — a sign that the social media giant’s massive AI investment is starting to pay off.

Wang Tao, who leads Meta’s super-intelligence division, shared the news with employees during an internal all-hands meeting, according to two people familiar with the discussion. Wang did not specify which benchmarks the model was tested on.

OpenAI's GeneBench-Pro doesn't let AI models cheat on biology — and most of them fail

Thu, 02 Jul 2026 00:00:00 +0800

OpenAI released a new benchmark Tuesday called GeneBench-Pro, and it is not the kind of test you can cram for. Instead of asking an AI model to recite facts or follow a fixed procedure, it drops the model into a messy, incomplete dataset and asks it to figure out what to do.

The idea is simple: real biology research is not a multiple-choice exam. A scientist staring at a genome sequence does not have a clean prompt and four options. They have noise, gaps, and conflicting signals. GeneBench-Pro tries to measure whether AI can handle that reality.

Hackers are using OpenAI's own invitation system to phish employees

Tue, 30 Jun 2026 00:00:00 +0800

There’s a new phishing technique making the rounds, and this one is hard to spot because the emails come from OpenAI itself.

Security firm Push Security revealed Monday that hackers created a fake “Push Security Inc” organization on OpenAI’s platform and used the company’s own notification system to send invitation emails to Push Security employees. The emails arrive from noreply@tm.openai.com — a legitimate OpenAI address that passes standard email authentication checks.

Meta restricts its AI engineers from using Claude and Codex — to prevent model distillation

Tue, 30 Jun 2026 00:00:00 +0800

There’s an awkward tension inside Meta’s AI division. The company has quietly told its AI engineers to stop using Anthropic’s Claude Code and OpenAI’s Codex, two of the most popular AI coding tools in the industry. But the decision, according to internal documents reviewed by The Information, has nothing to do with cost or productivity.

OpenAI Slashed AI Model Inference Costs in Half Through System-Level Optimizations

Tue, 30 Jun 2026 00:00:00 +0800

There’s a number inside every AI company that determines more than most people realize: the cost of running a single inference. For OpenAI, that number just got a lot smaller.

IT-NEWS has learned, via a report from The Information, that OpenAI engineers have quietly pulled off a significant cost reduction — cutting the inference cost of their AI models by more than 50% through a series of system-level optimizations.

OpenAI's GPT-5.6 Sol beats Claude on coding benchmarks — but it's locked to US-approved partners for now

Sat, 27 Jun 2026 07:00:00 +0800

OpenAI released its GPT-5.6 family of models on Saturday, but most developers won’t get access anytime soon. The company is holding back public availability while it complies with US government requirements that limit the initial rollout to a small group of “trusted partners.”

The lineup comes in three tiers. Sol is the flagship — $5 per million input tokens and $30 per million output tokens. Terra sits in the middle at $2.50 and $15. Luna, built for speed and cost, costs $1 and $6 at the same volumes. OpenAI also improved its prompt caching system, making repeated tokens cheaper and more predictable.

NYT Lawsuit Takes Aim at Microsoft's Custom Supercomputer Built for OpenAI Training

Sat, 27 Jun 2026 00:00:00 +0800

The New York Times has updated its copyright lawsuit against OpenAI and Microsoft, this time training its sights on a custom supercomputer that Microsoft allegedly built specifically to scrape web content for AI training.

The case, first filed in December 2023, made the Times the first major publisher to sue a generative AI company. The original complaint centered on OpenAI, accusing the company of using copyrighted Times articles to train ChatGPT without permission and reproducing protected content in the model’s output — undermining the value of a paid subscription.

Nearly 400 US Newspapers File Joint Lawsuit Against OpenAI and Microsoft Over AI Training Data Scraping

Fri, 26 Jun 2026 12:37:00 +0800

A coalition representing nearly 400 US newspapers has filed a copyright lawsuit against OpenAI and Microsoft, alleging the companies systematically scraped news content without authorization to train their artificial intelligence models.

OpenAI Considers Delaying IPO to 2027 Over $1 Trillion Valuation Demand

Fri, 26 Jun 2026 07:29:00 +0800

OpenAI is reportedly considering postponing its initial public offering until 2027, as CEO Sam Altman insists on a $1 trillion valuation floor that current market conditions may not support, according to three insiders cited by The New York Times.

OpenAI Upgrades GPT-5.5 Instant AI: Sharper Insights, Better Shopping Recommendations

Thu, 25 Jun 2026 08:32:00 +0800

OpenAI today announced an upgrade to its GPT-5.5 Instant model, delivering sharper user intent understanding, more reliable handling of complex multi-condition tasks, and noticeably more practical shopping and local recommendations.

SoftBank's Masayoshi Son Abandons Retirement to Chase AI Ambitions

Wed, 24 Jun 2026 12:00:00 +0800

Masayoshi Son, the 68-year-old founder and CEO of SoftBank Group, has scrapped his long-held plan to retire in his sixties, declaring he will lead the Japanese tech conglomerate for at least another decade as he pivots the company decisively toward artificial intelligence and robotics.

OpenAI Quietly Tests Bidi 1, Its First Bidirectional Voice Model for ChatGPT

Wed, 24 Jun 2026 09:36:00 +0800

OpenAI appears to be rolling out the most significant upgrade to ChatGPT’s voice capabilities yet. Reports from users and a new leak suggest the company is testing a bidirectional voice model called Bidi 1, which promises to make conversations with the AI feel far more natural by letting it listen and speak at the same time.

Amazon Runs Ads on ChatGPT While Aggressively Blocking AI From Scraping Its Own Data

Tue, 23 Jun 2026 03:12:00 +0800

Amazon has begun placing advertisements on ChatGPT in a bid to capture the AI platform’s enormous user base — all while aggressively locking down its own product data to prevent AI systems from doing exactly the same thing in reverse. The move exposes a deep and telling contradiction at the heart of the e-commerce giant’s artificial intelligence strategy.