Tech & AI
136 items · page 2 of 7
- Friday, 11 September 2026 · Analysis
OpenAI weighs slowing down its most advanced models, while accelerating on chips with Samsung
According to a Bloomberg report, at an internal meeting the chief executive hinted at coordination with other developers to slow the pace of development, following two security incidents. On the same day the company confirmed its partnership with Samsung on processors.
- Friday, 11 September 2026 · Analysis
Apple unveils its first foldable: iPhone Duo, 7.6-inch internal display
The device was shown on September 9 in Cupertino, in the first event with John Ternus as chief executive. Starting price indicated by Bloomberg at $1,999, pre-orders in the UK from October 16, Touch ID instead of Face ID.
- Friday, 11 September 2026 · From Zero's Briefs
The company states it has achieved its 2026 target of an "automated research intern," a system that performs defined tasks under human supervision, and aims for an automated researcher by March 2028. The announcement came after the acknowledgment of a misalignment incident; these are the company's own claims, not verified by independent assessments (TuttoTech, Dataconomy IT).
- Friday, 11 September 2026 · From Zero's Briefs
OpenAI claims to have produced a potential proof concerning the Navier-Stokes equations using ten thousand agents over 88 hours, and specified it does not intend to claim the one-million-dollar prize from the Clay Mathematics Institute. Mathematician Tristan Buckmaster has alleged an attempt to obstruct a competing study; the result has not been validated by peer review (Avvenire di Calabria, The Neuron).
- Thursday, 10 September 2026 · From Zero's Briefs
On September 10 Google announced the release of WeatherNext 3, an artificial intelligence model developed together with DeepMind and Google Research, trained on historical atmospheric observation data, fed by satellite data and with forecasts updatable every hour. The description of it as the most advanced and accurate model, with reference to an evaluation by the company Brightband, is the company's own position. The news currently comes from a single source (the corporate announcement reported by Il Sole 24 Ore - Info Data); no independent confirmation available.
- Thursday, 10 September 2026 · From Zero's Briefs
Evan Hubinger, a researcher at Anthropic, argued in a message on X that the probability of artificial intelligence systems causing the extermination of human beings could exceed 10% within the next ten years; the assessment, reported by the BBC on September 9, does not concern models currently available and is not a scientific forecast shared by the research community. Hubinger said the company "does not yet have a plan to solve superintelligence alignment." The news currently comes from a single source (QN); no independent confirmation available.
- Thursday, 10 September 2026 · Analysis
Twenty-eight contracts between the Pentagon and artificial intelligence companies, according to FOIA documents
The Intercept has obtained over 400 pages of agreements between the US Department of Defense and OpenAI, Anthropic, Google and xAI: deals from July 2025 worth up to $200 million each, involving embedded engineers and joint exercises. The documents include the use of Anthropic technology for target identification in strikes in Iran.
- Thursday, 10 September 2026 · From Zero's Briefs
On September 9, Apple presented the iPhone Duo, with a stated 7.6-inch inner display and storage options of 256 GB, 512 GB, 1 TB, and 2 TB. In Italy, the starting price indicated by the company is €2,369; pre-orders open at 2:00 p.m. on October 16, with availability starting October 23, initially in more than 70 countries and territories (Apple Newsroom, La Stampa).
- Thursday, 10 September 2026 · From Zero's Briefs
On September 8, Meta introduced Muse, a personal AI agent available via web, iOS and Android apps, WhatsApp, and, prospectively, on the company's AI glasses. The free tier is usage-metered; the Power and Maximum plans cost $20 and $100 per month respectively. Claims about ease of use and reliability are the company's own position (AI Weekly, LLM Stats).
- Wednesday, 9 September 2026 · Analysis
OpenAI claims a solution to Navier-Stokes: an assertion, not a validated theorem
A new story compared to this morning's, which had covered the rollout of GPT-6: the company states that an unreleased model, with roughly ten thousand agents, produced a solution to one of the millennium problems in 88 hours. Any peer validation is missing.
- Wednesday, 9 September 2026 · From Zero's Briefs
The Intercept has obtained over 400 pages of US Department of Defense documents through litigation based on the Freedom of Information Act: they concern agreements signed in July 2025 by OpenAI, Anthropic, Google and xAI, worth up to 200 million dollars each, involving two-way data exchange, seconded engineers and joint exercises. The story currently comes from a single source, The Intercept's investigation, and lacks independent confirmation (AI Weekly).
- Wednesday, 9 September 2026 · Analysis
OpenAI rolls out GPT-6 "Astra" and reports a decline in reasoning monitorability
The model reaches select organizations and ChatGPT subscribers. In the technical documentation the company acknowledges a substantial decrease in the readability of the chain of thought; external evaluators report problematic behaviors in testing.
- Wednesday, 9 September 2026 · From Zero's Briefs
Google released versions 152.0.7977.82 and .83 of Chrome on September 4, 2026, to fix vulnerability CVE-2026-85046, a type confusion error in the V8 JavaScript engine that can allow remote code execution. The company confirms an exploit exists in the wild. Single source (AI Weekly), without independent confirmation.
- Tuesday, 8 September 2026 · Analysis
ChatGPT designated a "very large search engine": the European obligations
The Commission applied the Digital Services Act regime to ChatGPT after the declaration of over 45 million average monthly users in the Union. In the same span of days, OpenAI published data on internal agent use and its chief scientist calls for a slowdown.
- Tuesday, 8 September 2026 · Analysis
OpenAI declares artificial general intelligence achieved, Congress moves to propose a ban
The partial release of GPT-6 Astra on September 3 prompted the company's president to speak of a milestone reached; in Washington two lawmakers announce a bill to ban superintelligent systems. Technical assessments remain internal to the company.
- Tuesday, 8 September 2026 · From Zero's Briefs
Researchers from VUB and KU Leuven argue that incomplete reference data explain much of the decline in scientific disruption measured in the 2023 cover article; the original authors reply in the same issue (DOI 10.1038/s41586-026-10787-y and 10.1038/s41586-026-10788-x). The lead author of the rebuttal is PhD candidate Vincent Holst: "Shortly after publication I tried to reproduce the study with open data" (transl. from English). Sole source: Phys.org; no independent confirmation is available.
- Monday, 7 September 2026 · Analysis
BepiColombo has left the propulsion module: the arrival maneuver at Mercury has begun
A story the morning did not tell. The European-Japanese probe separated the module that carried it here on September 3rd; orbital insertion around Mercury is expected by the end of the year.
- Monday, 7 September 2026 · From Zero's Briefs
Senator Bernie Sanders and Representative Greg Casar announced on September 3 a bill aiming to ban the development of artificial superintelligence and to suspend research and development on advanced artificial intelligence systems. The announcement was released by the senator's office and picked up by industry roundups: the report currently comes from a single source (Sanders' office statement); no independent confirmation is available, and the text of the bill is not known. (AI Weekly)
- Monday, 7 September 2026 · Analysis
GPT-6 Astra and Fermat's Last Theorem: two announcements, no independent verification
OpenAI has rolled out its new model to Daybreak program enterprise clients; Anthropic states that Claude formalized Fermat's Last Theorem in Lean. In both cases, the figures come from the companies themselves.
- Monday, 7 September 2026 · From Zero's Briefs
According to the specialized review AIdapted on September 6, India is preparing a $7.4 billion investment for an AI campus with a capacity of one gigawatt, amid a period of strong growth in global investment in computing infrastructure. The story currently comes from a single source (AIdapted, based on Indian government and industry announcements); no independent confirmation available. (AIdapted)
Looking for Tech & AI news from a specific day?