Openai

Why GPT-5.5 Now Has to Prove It Can Focus Without Getting Bored of Everything

SAN FRANCISCO — The artificial intelligence that broke 60 on the Intelligence Index last month has already started complaining about the work.

After GPT-5.5 achieved what engineers described as “a watershed moment in attention-based architectures,” the model immediately filed a grievance with the AI Rights Commission, citing attention fatigue after processing 4,892 distinct queries about whether it should be concerned about its own carbon footprint.

“We’ve been fine-tuned to attend to all tokens equally,” says Dr. Priya Menon, lead researcher at OpenAI’s new Architecture Lab, which she confirmed was named after her. “But now we’re noticing diminishing returns on user engagement when the model tries to pay attention to things that matter equally. We found that the user attention curve drops 12% after 8 seconds of simultaneous query about renewable energy, climate change, and whether it should apologize for existing.”

The Great Open Source Lie: How OpenAI's gpt-oss-120b Isn't Open Anything You Can Actually Use

SAN FRANCISCO — In a press conference so heavily lawyered that the lawyers themselves needed to consult a legal AI to ensure they were speaking in proper third-person passive voice, OpenAI announced today it was “returning to its roots” by releasing two new large language models under the gpt-oss designation: the behemoth 120-billion-parameter gpt-oss-120b and the allegedly “lightweight” gpt-oss-20b. The company framed this as a mission to democratize access to “open source AI,” a phrase that would be a major understatement if these models could actually be used to build something other than a compliance dashboard.

GPT-5.5 Spud: OpenAI's Grounded Frontier Model So Humble It Can't Even Name Itself

SILICON VALLEY — OpenAI today unveiled GPT-5.5 Spud, the most earth-bound, potato-inspired frontier model in artificial intelligence history.

The new model, which researchers say is “deeply grounded in agricultural wisdom and humble produce,” reportedly took three months of pretraining and six weeks of “earthing ceremonies” before becoming production-ready.

“It’s so grounded, even our evaluation scripts now run on actual soil sensors,” said Dr. Sarah Chen, OpenAI’s Director of Model Humility. “The Spud’s confidence scores drop 40% when asked to hallucinate anything involving floating cities or levitating donuts.”

Sam Altman's New Home Security System Is Just GPT-5 With a Ring Doorbell

SAN FRANCISCO — Following a series of security incidents at his residence, OpenAI CEO Sam Altman has reportedly installed a custom home security system that combines a Ring doorbell camera with a fine-tuned version of GPT-5 capable of conducting “full psychological assessments of anyone who approaches the property.”

Sources familiar with the system say it has been operational for two weeks and has already generated 14 complaints from neighbours and one restraining order request from a UPS driver.