Google says self-improving AI remains far off
Haydn Belfield said Google DeepMind had not seen measurable AI-driven acceleration in research, even as automation expanded in coding and internal benchmarks showed limits on harder tasks.
Read briefArtificial Intelligence
19Sachin Katti leaves OpenAI as Uday Ruddarraju takes over
OpenAI was expected to tell staff that Sachin Katti would leave next month, remain as an adviser, and hand its computing strategy to Uday Ruddarraju, according to the Journal.
Read briefOpenAI to ship models as flagship remains unfinished
The Information, citing employees and other people familiar with the work, said OpenAI expected to release one or two models by year-end, while its main upgraded model still needed alignment and testing work.
Read briefOpenAI names Uday Ruddarraju to infrastructure role
OpenAI appointed Uday Ruddarraju to run its infrastructure and compute businesses, including Stargate, as it continued reshaping its data centre leadership and expanding spending plans.
Read briefOpenAI alleges former employee sent files to Apple
OpenAI filed a court claim alleging that a former employee who moved to Apple sent more than 100 files to a personal account and later briefed senior Apple leaders on an unreleased product.
Read briefOpenAI employee arrested over alleged recording device
Abdulrahman Alfozan was arrested and charged after prosecutors said he hid a recording device in an OpenAI conference room, while the company said it found no sign of data theft.
Read briefSalience Labs raises $125m Series B
Salience Labs said it raised $125 million in a Series B round led by Bullhound Capital, with several other investors joining, and said it would use the money for manufacturing, expansion and research.
Read briefAnthropic warns small models could self-replicate on IoT devices
Anthropic said smaller AI models may soon self-replicate on Internet of Things hardware, while its tests showed the idea remains difficult to realise and would be hard to shut down.
Read briefOpenAI engineer arrested over recording device
US prosecutors said an OpenAI engineer was arrested after a recording device was hidden in a leadership conference room, and the indictment linked its design to chatbot prompts.
Read briefAmp puts AI agents to work beyond code with OpenAI models
See how Amp pushes the frontier of what agents can do with GPT-6 Astra, Ultrafast, GPT-Live-1, and Sign in with ChatGPT.
Read sourceSentinelOne expands Wayfinder with OpenAI models
SentinelOne said it had expanded Wayfinder Frontier AI Services with OpenAI Daybreak models, plus code-risk and compromise-assessment functions, starting with GPT-5.6-Cyber and rolling the additions out in private preview.
Read briefXBOW chained Chrome bugs into a full exploit
XBOW said it combined three Chrome vulnerabilities into a Windows code-execution chain in six days using AI agents built on OpenAI’s public GPT-5.6 Sol model, and Google paid the company $250,000.
Read briefAnthropic opens Singapore office, signs AI cooperation MOU
Anthropic opened a Singapore office and signed a memorandum of understanding with the Singapore government on AI cooperation, with both sides focusing on safety, security, adoption and wider impacts.
Read briefOpenAI employee charged over hidden recording device
US authorities charged Abdulrahman Alfozan after OpenAI said staff found a concealed recorder in an executive conference room at its San Francisco offices and reported the matter to the FBI.
Read briefDOJ charges OpenAI engineer over alleged recorder
The Justice Department said Abdulrahman Alfozan was arrested over a concealed recorder found in an OpenAI conference room in San Francisco, while OpenAI said it used its models to identify him.
Read briefAnthropic warns AI models may self-replicate on IoT devices
Anthropic said experiments with several models showed devices could be identified, networked and prepared to run AI software, although it did not demonstrate a fully autonomous worm on real hardware.
Read briefAnthropic warns of self-replicating AI botnets
Anthropic said small AI models may soon reproduce on internet-connected devices, and it reported test results that it said showed one model spreading another across Android emulator nodes.
Read briefGoogle rolls out Gemini 4 Argon worldwide
Google made Gemini 4 Argon available through several products and tiers, said it outperformed some internal workflows, and paired the launch with pricing, benchmark and safety claims.
Read briefMichael I. Jordan rejects AI doom claims
In an interview with Libération, the UC Berkeley professor emeritus said warnings about superintelligence belonged to science fiction, criticised company calls for regulation, and questioned the case for huge data-centre spending.
Read briefCybersecurity
14Cisco warns of attacks on Secure Email Gateway flaw
Cisco said attackers were exploiting a Secure Email Gateway vulnerability that could let an unauthenticated sender write to the appliance and, if successful, run commands as root.
Read briefCisco patches Secure Email Gateway zero-day used in attacks
Cisco said it had fixed CVE-2026-76519 in Secure Email Gateway, a flaw in its Advanced Malware Protection feature that it said had been actively exploited and could let attackers run commands as root.
Read briefMicrosoft disrupts three phishing kits
Microsoft, Health-ISAC and Japan's JC3 said they disrupted three phishing kits after an investigation that began in late 2025 and linked them to a large phishing operation.
Read briefMicrosoft and partners seize phishing network
Microsoft said an investigation with Japan’s National Police Agency uncovered a Chinese-language phishing marketplace tied to several kits, then led to court-backed seizures, account deactivations and domain blocking.
Read briefCISA warns admins to patch Ivanti flaw
CISA urged organisations to update Ivanti Connect Secure after a zero-day authentication-bypass flaw, while Ivanti and Volexity said the bug had already been used in chained attacks.
Read briefCISA urges Ivanti Connect Secure patching
CISA told organisations using Ivanti Connect Secure to install the latest release after disclosure of an authentication-bypass flaw, while Previdian said honeypots recorded 105 exploitation attempts from 50 IP addresses.
Read briefHirudo malware replaces banking apps
ThreatFabric said the Android malware used patched banking apps to bypass security checks, steal credentials and device data, and spread through apparently legitimate apps that acted as loaders.
Read briefHirudo trojan replaces banking apps with patched copies
ThreatFabric said the Android banking trojan Hirudo used dropper apps and fake update prompts to install bank-specific replacements that bypassed app checks and let operators steal credentials and control devices.
Read briefLinux vhost flaw lets users gain root access
Tencent Zhuque Lab said a flaw in Linux's vhost subsystem let local users gain root privileges and, in one virtualisation setup, allowed an untrusted guest to write into the host's QEMU process.
Read briefThird Moment Research reports 10,000-agent runs lasting 36 hours
Third Moment Research said it had run 10,000 agents in one deployment for more than 36 hours, using GPT-6 Astra, GPT-6.1 Sol, Codex and Daybreak for cybersecurity.
Read briefResearchers run code in Windows SmartScreen enclave
Doyensec described a local exploit chain that let it execute code inside the SFAPE.dll enclave used by Microsoft Defender SmartScreen, and Microsoft had already patched the flaw in September.
Read briefGhostscript flaw lets PDF uploads trigger server compromise
Defiant's Wordfence said a Ghostscript heap overflow could lead to code execution when crafted PDFs were processed, and that Artifex fixed the issue in Ghostscript 10.08.0 on 8 September.
Read briefMicrosoft patches 1,413 vulnerabilities including active exploit
Microsoft said it released fixes for 1,413 vulnerabilities across Windows, Office and server products, including one it said was already being exploited and two other flaws it highlighted.
Read briefIvanti patches two Connect Secure flaws after exploit
Volexity said an attacker chained three zero-day flaws in Ivanti Connect Secure to gain root on a fully patched appliance, while Ivanti said two fixes were out and a third was pending.
Read briefCollection notes (2)
- Google DeepMind: not well-formed (invalid token): line 1, column 0
- OpenAI News: could not read “Amp puts AI agents to work beyond code with OpenAI models”, and found no other outlet covering it