Meta and Microsoft cut back on Claude use
Meta and Microsoft reduced internal use of Anthropic’s Claude, with Microsoft shifting staff towards its own tools and Meta lowering Claude Code usage while promoting its own products.
Read briefArtificial Intelligence
15Meta and Microsoft cut internal use of Claude
The Information reported that Meta Platforms and Microsoft were reducing staff use of Anthropic’s Claude and moving workers towards their own tools, while Meta’s internal Claude Code usage fell and Microsoft’s budget estimate dropped.
Read briefNew York City Council hears AI safety testimony
The New York City Council was due to hear from AI company representatives and former Anthropic researcher Jacob Coxon as it weighed guardrail proposals and a subpoena dispute involving SpaceXAI.
Read briefOpenAI plans EU ChatGPT watermarks and optional API access
OpenAI said it would switch on invisible text watermarking for ChatGPT and Codex in the EU, while keeping API watermarking optional worldwide and limiting detector access at first.
Read briefPentagon stops using Anthropic AI tools
A US defence official told the BBC the Pentagon had ceased using Anthropic products after months of dispute over military access, safety guardrails and the company’s national-security designation.
Read briefTo comply with the EU AI Act, OpenAI plans to add text watermarking for ChatGPT and Codex users in the EU and an…
OpenAI : To comply with the EU AI Act, OpenAI plans to add text watermarking for ChatGPT and Codex users in the EU and an opt-in setting for API customers globally — - Starting today, API customers globally will be able to opt in to text watermarking for select models.
Read sourceAnthropic donations hit $540 million in 2025
Documents shared with potential investors said Anthropic topped up employee charity donations with company shares, lifting 2025 gifts to $540 million and costing the company more than $660 million in six months.
Read briefOur approach to EU text provenance rules
How OpenAI is approaching text watermarking under EU rules. Learn where watermarks apply, how detection works, and why access starts with researchers.
Read sourceFormer AI lab researchers explain why they quit
The article gathered interviews with former OpenAI, Anthropic and Google DeepMind researchers about leaving their labs, focusing on safety fears, internal frustrations and disagreement over how to respond.
Read briefImport AI surveys swarm scaling and AI science economy
The newsletter discussed swarm scaling, polling on AI self-regulation, Google DeepMind’s biology watermarking work, a lab benchmark, and a paper on markets for automated science.
Read briefApple plans tighter macOS Full Disk Access controls
Apple said it would change macOS Full Disk Access so broad system permissions required explicit user action, after reporting on AI agents, Meta's Muse and other Mac app vulnerabilities.
Read briefOpenAI to test visual ads in ChatGPT images
OpenAI said it would begin testing labelled visual adverts inside ChatGPT image generation in the US this month, while also adding measurement and brand-safety tools for advertisers.
Read briefOpenAI plans image ads and broader measurement in ChatGPT
OpenAI said it would test image ads in ChatGPT in the US and widen measurement, attribution and brand-suitability partnerships as it tried to answer advertiser concerns about spend and inventory.
Read briefTrump announces Super Intelligence Force
Donald Trump announced a White House Super Intelligence Force to coordinate outreach on artificial intelligence, with several officials named to lead it and reporting lines set to the president and his chief of staff.
Read briefAnthropic asks Claude users for voice data
Anthropic added a Claude voice prompt that asked users whether the company could use voice recordings and voice chat data to improve its AI models, with separate privacy controls for different kinds of training.
Read briefCybersecurity
10Microsoft patches Exchange flaw exposing mailboxes
Microsoft released out-of-band fixes for an Exchange Server vulnerability that could let an authenticated attacker read other users’ mailboxes within the same organisation, while Exchange Online had already been fixed.
Read briefDenmark registry breach exposed 8.8 million records
Denmark’s Central Population Register said a September incident exposed personal data on about 8.8 million registered people, and authorities said police were investigating how the access was abused.
Read briefDell warns of root-level flaw in System Update tool
Dell said a critical path-traversal bug in its System Update command-line tool could let remote attackers gain root access on unpatched machines, and it told customers to move to version 2.3.0.0 or later.
Read briefWeekly recap covers NetScaler flaws and AI leaks
The weekly roundup reported exploited flaws in Citrix NetScaler and Fortinet FortiMail, arrests tied to ShinyHunters and KillSec, state-linked malware activity, and screenshots exposed by AI coding agents.
Read briefUS announces arrest of Ploutus developer
The Justice Department said the alleged Ploutus developer was arrested and linked him to ATM jackpotting attacks, laundering, and charges filed in Nebraska.
Read briefCling botnet exploited Realtek Jungle SDK flaw
Nozomi Networks said it saw exploitation of CVE-2021-35394 in Realtek Jungle SDK from around 5 September 2026, with some attempts delivering Cling and using STUN traffic for command and control.
Read briefGoogle suspends OSS VRP submissions
Google temporarily stopped new product vulnerability submissions to its open-source rewards programme after a rise in automated reports, while other report types and existing cases remained open.
Read briefAttackers exploit Rejetto HFS session-forgery flaw
VulnCheck said attempts were under way against a Rejetto HTTP File Server flaw that could let an attacker forge administrator sessions and reach code execution.
Read briefCitrix patches NetScaler flaw after targeted attacks
Citrix said it fixed CVE-2026-88779 in NetScaler ADC and NetScaler Gateway after targeted attacks, and CISA added the flaw to its federal patch list.
Read briefCitrix patches exploited NetScaler SAML flaw
Citrix released emergency updates for CVE-2026-88779 after targeted attacks on NetScaler appliances using SAML authentication, while administrators and researchers examined whether the flaw could also enable code execution.
Read briefCollection notes (2)
- Techmeme: could not read “To comply with the EU AI Act, OpenAI plans to add text watermarking for ChatGPT and Codex users in the EU and an opt-in setting for API customers globally (OpenAI)”, and the coverage found instead turned out to be a different story
- OpenAI News: could not read “Our approach to EU text provenance rules”, and found no other outlet covering it