In today’s cybersecurity news…

Anthropic tells G7 to cooperate

The Financial Times’ sources say Anthropic CEO Dario Amodei used a G7 meeting to push for more international cooperation on AI after the US administration blocked exports of Anthropic’s latest models, raising concerns that the U.S. could limit allies’ access to cutting-edge AI. OpenAI CEO Sam Altman and Google DeepMind CEO Demis Hassabis reportedly backed the idea. France’s Emmanuel Macron warned that countries need to know they won’t be cut off from AI systems that are important to their economies and national security. (Financial Times)

Fortinet VPN leak exposes credentials

A new leak called FortiBleed appears to have exposed usernames and plaintext passwords for roughly 74,000 Fortinet VPN and firewall devices used by organizations worldwide. Security researcher Bob Diachenko uncovered the dataset, and independent analysis by cybersecurity researcher Kevin Beaumont suggests the credentials are authentic and may have come from exported Fortinet configurations, but it’s unclear how the data was taken. The leak may cover roughly half of all internet-accessible Fortinet firewalls and could give attackers a way to target companies, government agencies, and critical infrastructure operators. (BleepingComputer)

Crypto Clipper abuses reviews, narrators, and comments

Here’s a reminder to give popularity metrics online a second look. Researchers at Check Point found a malware campaign using fake reviews, inflated download counts, AI-generated YouTube tutorials, and even coordinated comments on VirusTotal to make a cryptocurrency-stealing tool look legitimate. The malware swaps copied crypto wallet addresses for attacker-controlled ones, and shows how attackers keep building fake reputations across trusted platforms to trick people into downloading malicious software. (The Hacker News)

Estonia assigns personal ID numbers to AI agents

Estonia is trying to get ahead of a problem other governments may face later. The country plans to become the first nation to give AI assistants their own digital ID numbers for better control over what AI agents can access and what actions they’re authorized to take, on behalf of people and businesses. This is a broader effort to address the legal and security challenges of increasingly autonomous AI systems as governments wrangle with how the tools should operate in the real world. (Bloomberg)

Huge thanks to our episode sponsor, ThreatLocker

Telegram challenges India’s block 

Telegram has asked the Delhi High Court to overturn its temporary block in India, which is imposed until June 22nd after authorities alleged cheating rings were using Telegram to share leaked materials related to a national medical entrance exam. The block affects more than 150 million users in India and includes both telecom-level restrictions and removal from app stores. Telegram argues the measure is disproportionate because it shuts down the entire platform to address the actions of a small number of bad actors. The court will now decide whether the ban should remain in place. (The Next Web)

Microsoft confirms RoguePlanet zero-day

Windows Defender has another zero-day on its hands. Security researcher Chaotic Eclipse, also known as Nightmare Eclipse, publicly released a proof-of-concept exploit called RoguePlanet that can grant SYSTEM-level privileges on vulnerable Windows machines, and Microsoft has now confirmed the flaw and says a patch is in development. It’s the fourth Defender vulnerability disclosed by Chaotic Eclipse in recent months. (The Hacker News)

Kodak confirms stolen records

Kodak confirmed a data breach after the hacking group ShinyHunters claimed it stole 2.2 million records, including customer personal information and internal company data. The company says the incident was contained and didn’t affect operations, but ShinyHunters has threatened to leak the data if Kodak doesn’t  respond by June 18th. The breach is the latest in a series of attacks linked to the group. (Security Magazine)

Google shifts IP address use for personalization 

Google is making a big change to how it targets ads in Europe. Starting in August, the company will start using IP addresses for ad measurement and personalization in the UK, EU, and Switzerland, raising privacy questions because IP addresses are considered personal data under GDPR and can be used as part of device fingerprinting. This comes less than two years after Google dropped its longstanding opposition to fingerprinting, and could raise new issues with regulators who argue this kind of cross-service tracking needs explicit user consent. (BleepingComputer)