Google can now use your Gmail data for ads: How to turn it off
Google’s Gmail settings can allow email data to inform ad personalization. Users can turn off the relevant personalization and smart-feature settings to limit that use. (proton.me)
Wiz’s investigation tracks remediation and attempted exploitation of two PHP vulnerabilities disclosed as PHP2Shell on October 21, 2026. Its telemetry shows the number of distinct vulnerable, internet-exposed machines fell about 67% in the following days, while organizations with at least one exposed vulnerable instance declined by 47.9%. The share of vulnerable machines running PHP 7.3 or earlier also dropped from 8.3% to 5.7% in five days. A week after disclosure, fewer than 53% of initially affected organizations remained vulnerable. These trends are consistent with patching, upgrades, machine retirement, or access restrictions, though the data cannot distinguish their individual effects.
Wiz observed exploit attempts but no completed PHP2Shell chain in its own telemetry. An older exploit-kit variant was blocked by request-size limits or failed to trigger the required memory exhaustion. A newer variant reliably exhausted PHP memory, but its next step - leaking memory addresses - failed in every captured run, and attackers did not proceed to code execution. The chain depends on successive requests reaching the same Apache worker and on memory layouts that vary with PHP builds, extensions, and configuration, making generic exploitation unreliable. Some PHP2Shell-labeled tools instead used session poisoning combined with a separate local-file-inclusion flaw, so they did not exploit the disclosed vulnerabilities on their own. Wiz notes that public honeypot evidence from another researcher did show successful exploitation under a favorable single-worker configuration, but does not establish reliability across deployments.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Google’s Gmail settings can allow email data to inform ad personalization. Users can turn off the relevant personalization and smart-feature settings to limit that use. (proton.me)
A viral post claims Claude Code found TikTok, Instagram, Facebook, and Snapchat recording conversations in the background, but provides no verifiable evidence. The allegations are unsubstantiated; users can review microphone permissions and background activity in their phone settings. (twitter.com)
Discover-Tech, previously linked to zero-click spyware delivery, was the top bidder for Starlink ad impressions in 105 countries, prompting researchers to warn of a possible surveillance operation targeting users. The activity may focus on countries where Starlink is banned or restricted, though the findings do not establish who is behind the effort or its exact purpose. (gizmodo.com)
A critical memory-corruption flaw in TianoCore EDK II UEFI firmware lets attackers run code during startup, before defenses such as Secure Boot take effect. The flaw affects firmware used by most server vendors and data centers, and attackers could use it to install persistent boot-time malware. (arstechnica.com)
Anthropic says its unreleased Claude model found and chained critical vulnerabilities in widely used software, including a one-click Chrome exploit that achieved remote code execution. The company reported the flaws to vendors; Google assigned and patched 62 related Chrome CVEs. (anthropic.com)
Tailscale is investigating security vulnerabilities affecting Tailscale SSH and has shared an update on its findings and remediation efforts. The company is also providing guidance for customers. (tailscale.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.