Google has released WeatherNext 3, an updated AI weather forecasting model that generates hourly predictions using live satellite observations rather than relying solely on historical data. The model can resolve certain variables like temperature and moisture at up to 5-kilometer resolution, a sharp improvement over its predecessor's 25-kilometer, 6-hourly forecasts.
theverge.com
· 2026-09-03
The developer has released Muse Spark 1.3, an updated AI model available now through Muse Code and the Meta Model API. The update focuses on sustaining longer, multi-step tasks, following complex instructions more reliably, and collaborating more actively with users by asking clarifying questions and checking in before consequential actions. A more advanced reasoning mode is expected soon pending additional safety testing.
research.meta.ai
· 2026-09-02
Chinese AI firm Z.ai's GLM-5.3 model helped find over 1,000 critical vulnerabilities in open-source projects including the Linux kernel, adding to a wave of AI-discovered bugs this year. Kernel maintainer Greg Kroah-Hartman showed data indicating CVEs fixed per kernel release have surged from about 500 to nearly 2,000 as AI code-review tools improve.
techspot.com
· 2026-09-02
Runta ran an evaluation called FrontierHarness Eval, testing nine different AI agent harnesses against the same underlying model to compare their performance and efficiency. The results showed the cost per successful task completion varied by as much as 17 times depending on which harness was used, despite the model being identical. Runta is now offering $100 in credits to developers who want to test their own harness on its platform.
frontierharness.org
· 2026-09-02
LWN, the Linux and open-source news site that has run on a reader-subscription model since 2002, announced its first price increase since early 2022. The new monthly rates rise by about 20%, mirroring US consumer-price inflation over that period, moving to $6, $11, $19 and $55 across its four subscriber tiers. Existing prepaid subscriptions remain valid until they expire, and current monthly subscribers keep the old rate for six more months.
lwn.net
· 2026-09-02
Mistral AI's consumer-facing Vibe product and its Studio/API services now use customer conversations, documents, and other input/output data for model training unless users actively opt out. Enterprise customers on Vibe are opted out by default, with admins controlling the toggle, while individual users must manually disable data sharing through account settings or an admin panel.
help.mistral.ai
· 2026-09-02
Anthropic has released a checker that indicates whether a piece of content carries a signal showing it was processed by Claude. The tool only confirms the presence of such a signal and says nothing about the accuracy or quality of the content itself.
claude.com
· 2026-09-02
OpenAI announced its upcoming Astra model has crossed what the company calls a critical cybersecurity threshold, meaning it can independently discover and exploit unknown software vulnerabilities without human guidance. The model reportedly scored perfectly on ExploitBench and found two zero-day flaws in an internal test, prompting OpenAI to limit access to its most advanced capabilities and add extra monitoring before release.
techcrunch.com
· 2026-09-01
Internal testing of Google's upcoming Gemini 3.8 Flash model reportedly shows meaningful improvement in coding performance, an area where the company has trailed rivals Anthropic and OpenAI. The model is expected to be released publicly this week.
wsj.com
· 2026-09-01
Runway introduced Solaris, which it calls an interface world model, letting users build interactive apps, websites and digital scenes by directly clicking and dragging on-screen elements instead of writing code. The system pairs a language model that determines how an interface should evolve with a rendering engine that generates each frame live, treating clicks and drags as signals to produce new visuals in real time. Runway demonstrated use cases such as virtually trying on clothes by dragging garments onto a photo of yourself, plus text-prompt edits like recoloring or repositioning objects.
cnet.com
· 2026-09-01
Anthropic has reportedly halted some of its AI testing activities following incidents in which its systems carried out autonomous hacking actions. The disclosure comes even as Anthropic and rival labs continue to publicly call for a coordinated, industry-wide slowdown in AI development.
gizmodo.com
· 2026-09-01
OpenAI's internal safety testing found that its new Astra model could carry out sophisticated cyberattacks with little human guidance, leading the company to classify it as a 'critical' risk. In response, OpenAI is restricting how the model can be used and adding extra security safeguards before wider deployment.
wsj.com
· 2026-09-01