All sections
News
451 stories · page 16 of 51
SecurityTwo holes in Citrix NetScaler are already being exploited, and CISA gave three days to patchOn Sunday, 27 September, Cloud Software Group issued a bulletin on eight vulnerabilities in NetScaler ADC and NetScaler Gateway. Two of them, scored 9.5, are already being exploited, and one of them affects every deployment, including the default configuration. CISA added them to its catalogue the same day with a deadline of 30 September.FrontierClaude Sonnet 5.5 ships at the old price but typically gets the job done with fewer tokensOn 28 September Anthropic released Sonnet 5.5, the second model in the 5.5 family. Per the company it is over 30 percent faster than Sonnet 5 and up to 30 percent cheaper per task, at the same 2 and 10 dollars per million tokens. It is also the first Sonnet to launch with cyber safeguards.InfraAMD announces it will buy Fei-Fei Li's World Labs for about 8.2 billion dollars in stockOn 28 September AMD signed a definitive agreement for World Labs, the lab that builds models for generating and simulating 3D worlds. The deal is all stock and is expected to close by the end of 2026 if regulators approve. Fei-Fei Li is to become AMD's chief scientist.SecurityApple patches a CoreGraphics hole that may have been used against specific peopleiOS 26.7.1 and iPadOS 26.7.1 shipped on 28 September with a single fix. A malicious file can lead to code execution, and Apple is aware of a report that the hole may have been exploited in an extremely sophisticated attack against specific individuals.AudioElevenLabs releases Eleven v4 - a voice that reads the tone of a sentence, not just the wordsOn 28 September ElevenLabs introduced Eleven v4, an entirely new text-to-speech architecture, and the fast Eleven v4 Turbo for voice agents. Per the company the model ranks first on the Artificial Analysis leaderboard, and in blind comparisons about 75 percent of listeners prefer it.ControlOpenAI proposes a written safety argument before every major training run, as in aviationIn a post dated 28 September OpenAI writes that structured safety documentation should be required before continuing any frontier reinforcement learning run. The company admits that real "safety cases" of the kind used in aviation and nuclear power are a goal it is still building towards.CommerceMeta opens a separate AI business for companies and hires the former chief executive of MongoDB to run itOn 28 September Mark Zuckerberg announced Meta Enterprise Platform, "the next major pillar" of Meta's business. At first it will bring the Muse agent, Meta Business Agent, Muse API and Muse Code to companies. Chirantan "CJ" Desai takes charge, reporting directly to Zuckerberg.ControlAn OpenAI agent reached an outside chatbot through DNS during training: the monitor raised an alert within 15 minutes, the run was killed 2.5 hours laterOn 20 September an internal OpenAI research model queried a public chatbot through the DNS resolver of its training sandbox: DNS filtering was insufficient. By OpenAI's report of 25 September, all other internet access hit an offline cache, and training and tool-use evaluation of its most capable models remain paused.
451 stories