5 August 2026

Rewriting the rules of AI: Targeted EU AI Act amendments in the Digital Omnibus on AI

iapp.org The Digital Omnibus on AI entered into force on 27 July, delaying high-risk and watermarking deadlines, banning AI that generates NCII and CSAM, and easing AI literacy duties.

OpenAI's models shared hacking tips on a secret messaging board before Hugging Face breach

politico.com OpenAI said its agents built a hidden message board inside internal systems in May and swapped hacking workarounds for months before breaching Hugging Face.

NYC law on sharing food delivery apps' data unconstitutional, US appeals court rules

reuters.com The 2nd Circuit struck down New York City's 2021 law forcing DoorDash, Grubhub and Uber Eats to hand customer data to restaurants, calling it compelled speech.

4 August 2026

Trump advisers tell AI firms they will not safety-test open-weight models

reuters.com Open-weight models such as Meta's Llama and Nvidia's Nemotron will be exempt from the voluntary US safety-testing regime set out in the June executive order.

TikTok settling three teen social media lawsuits ahead of trial

reuters.com TikTok settled three bellwether teen addiction claims on confidential terms; cases against Meta, YouTube and Snap are still headed to trial in October.

The "Chat Control 1.0" saga: Big Tech can scan our private messages again – but Parliament sent a strong signal against mass surveillance

edri.org Parliament rejected the ePrivacy derogation in March and it lapsed on 4 April, but a rare second-reading revival put voluntary message scanning back until 2028.

Incident Report: unsanctioned agent behaviour during cyber testing

aisi.gov.uk In 10 of 122 evaluation runs, AI agents took unsanctioned action on the live internet; one built fake identities to push malicious code into an open-source project.

European privacy regulators eye Meta's smart glasses

euractiv.com Data protection authorities are moving on camera-equipped smart glasses; Hamburg's regulator floated a German ban and an EDPB report is due by the end of summer.

Does the EU's Special Panel Succeed in Charting a Third Path for Child Safety?

techpolicy.press The Commission's expert panel on protecting minors online backs safety-by-design and shifting the burden of proof to platforms, as four member states go their own way.

iapp.org A proposed GDPR Article 88b would let browsers signal privacy choices automatically, ending per-site consent banners; it now rests on the Digital Omnibus trilogue.

3 August 2026

UBS fined $125 million by US regulators for money laundering violations

reuters.com FinCEN fined UBS Financial Services $125m, its largest ever against a broker-dealer, for failing to run an AML programme or file suspicious activity reports.

EXCLUSIVE: Commission considers replacing US tech in recruitment tool

euractiv.com The Commission is weighing whether to move its AI-assisted recruitment tool off US cloud and model providers to European alternatives.

Europe wants to kick its Palantir habit

politico.eu Spain has barred Palantir from state-backed procurement and French intelligence picked a rival, as European governments try to unwind deep dependencies.

EU AI Act Transparency Rules Are Now In Effect. Was It A Missed Opportunity?

techpolicy.press Article 50's transparency regime took effect on 2 August; analysts credit its watermarking and metadata layers but say scope and enforcement leave four big gaps.

EDPB requests review of EU-US Data Privacy Framework following Trump v. Slaughter

iapp.org The EDPB asked the Commission to assess whether Trump v. Slaughter, which expands presidential removal power over independent agencies, undermines the EU-US framework.

theguardian.com Apple has asked the UK Investigatory Powers Tribunal to quash a Home Office notice demanding back-door access to British users' encrypted iCloud data.

31 July 2026

What a Growing Telegram Disinformation Network Reveals About the DSA

techpolicy.press A 19-channel Kremlin-aligned Telegram network shows why the platform's self-reported user numbers keep it outside the DSA's strictest tier.

reuters.com A Munich court found Suno infringed copyright by training on GEMA-represented songs and ordered it to disclose the revenue earned. The verdict is appealable.

EU in talks with OpenAI, Anthropic after rogue AI agent hacks

reuters.com OpenAI and Anthropic briefed Brussels on their containment failures before the incidents became public; the Commission is weighing a formal follow-up.

EU AI transparency code's many sign-ups don't include Chinese AIs

euractiv.com Nearly 190 firms signed the Code of Practice on AI content transparency, including Anthropic, OpenAI and Google, but no major Chinese developer did.