<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>The Raw Logs — #security</title><link>https://therawlogs.com/tag/security</link><description>Top takes tagged #security.</description><language>en</language><lastBuildDate>Tue, 08 Sep 2026 16:06:29 GMT</lastBuildDate>    <item><title>#security — OPSWAT Launches Security Inspector to Block Poisoned AI Files</title><link>https://therawlogs.com/take/20-3</link><guid>https://therawlogs.com/take/20-3</guid><pubDate>Mon, 07 Sep 2026 16:25:16 GMT</pubDate><description><![CDATA[OPSWAT Launches Security Inspector to Block Poisoned AI Files: OPSWAT released its AI Inspector at GISEC 2026, scanning incoming files to block hidden prompt injection attacks before data ingestion. Scan and sanitize all incoming files before adding data to search databases. [#security]]]></description></item>    <item><title>#security — EU weighs Cybersecurity Act 2.0 bans across supply-chain tiers</title><link>https://therawlogs.com/take/17-3</link><guid>https://therawlogs.com/take/17-3</guid><pubDate>Fri, 04 Sep 2026 16:38:17 GMT</pubDate><description><![CDATA[EU weighs Cybersecurity Act 2.0 bans across supply-chain tiers: The EU has begun reviewing Cybersecurity Act 2.0, banning tech that uses parts from high-risk vendors across all supply tiers. Relying on the primary vendor alone fails when sub-chips are unvetted. Companies must audit every component, down to the raw circuit board. [#security]]]></description></item>    <item><title>#security — Palo Alto Unit 42 Discloses AI Attack Compressing a Two-Week Breach to 10 Hours</title><link>https://therawlogs.com/take/16-2</link><guid>https://therawlogs.com/take/16-2</guid><pubDate>Thu, 03 Sep 2026 16:50:11 GMT</pubDate><description><![CDATA[Palo Alto Unit 42 Discloses AI Attack Compressing a Two-Week Breach to 10 Hours: Unit 42 confirmed an attacker used frontier AI to shrink a two‑week corporate network breach to ten hours. Human security teams cannot match automated attack scripts. Defense systems must revoke user tokens and isolate machines within seconds. [#security]]]></description></item>    <item><title>#security — Google Cloud aims at high‑risk workflows using regulated legal platforms.</title><link>https://therawlogs.com/take/10-3</link><guid>https://therawlogs.com/take/10-3</guid><pubDate>Wed, 26 Aug 2026 16:15:43 GMT</pubDate><description><![CDATA[Google Cloud aims at high‑risk workflows using regulated legal platforms.: Google Cloud launched purpose-built governed enterprise platforms with deterministic citation verification. Unconstrained general models hallucinate citations and run up token bills. Governed domain tools enforce exact document references and fixed costs for legal workflows. [#security]]]></description></item>    <item><title>#security — Continuous attack-surface remediation shortens software patch windows to four hours.</title><link>https://therawlogs.com/take/9-3</link><guid>https://therawlogs.com/take/9-3</guid><pubDate>Tue, 25 Aug 2026 16:40:45 GMT</pubDate><description><![CDATA[Continuous attack-surface remediation shortens software patch windows to four hours.: Automated security tools now find external code vulnerabilities, write tests, and submit patches in 4 hours instead of the standard 60-day manual cycle. Connecting scanners directly to code repositories fixes exposed flaws before attackers exploit them. [#security]]]></description></item>    <item><title>#security — Outbound Proxy Rules for Containing Autonomous AI Agents.</title><link>https://therawlogs.com/take/8-1</link><guid>https://therawlogs.com/take/8-1</guid><pubDate>Mon, 24 Aug 2026 16:49:40 GMT</pubDate><description><![CDATA[Outbound Proxy Rules for Containing Autonomous AI Agents.: CSIS advises using an outbound proxy to stop agents from leaking data. Firewalls check only incoming traffic, so agents can bypass them by making outbound calls from your servers. Send all agent traffic through a proxy that blocks unauthorized destinations. [#security]]]></description></item>    <item><title>#security — Autonomous AI cyber attacks and new AI safety policies focus on computing power.</title><link>https://therawlogs.com/take/7-1</link><guid>https://therawlogs.com/take/7-1</guid><pubDate>Sun, 23 Aug 2026 14:58:19 GMT</pubDate><description><![CDATA[Autonomous AI cyber attacks and new AI safety policies focus on computing power.: AI can attack network systems, but safety rules only cover hardware, not open‑source models. Tools can change firewall settings quickly when a scan is detected. Banning open models would prevent companies from using their own tools. [#security]]]></description></item>    <item><title>#security — Regulators mandate AI transparency audits as agents run 4+ hours</title><link>https://therawlogs.com/take/6-4</link><guid>https://therawlogs.com/take/6-4</guid><pubDate>Sat, 22 Aug 2026 16:49:15 GMT</pubDate><description><![CDATA[Regulators mandate AI transparency audits as agents run 4+ hours: US, EU, and Singapore regulators require firms to disclose AI workings and run audits; using AI without an audit can lead to bans. New agents keep state and run over four hours; older agents stop after 10–15 minutes. You can now give AI tasks that run for hours with checkpoints. [#security]]]></description></item>    <item><title>#security — Required safety rules and required group shutoff switches.</title><link>https://therawlogs.com/take/5-5</link><guid>https://therawlogs.com/take/5-5</guid><pubDate>Fri, 21 Aug 2026 16:29:25 GMT</pubDate><description><![CDATA[Required safety rules and required group shutoff switches.: California lawmakers passed AI legislation that requires hardware emergency kill switches and independent safety audits for large training clusters. AI oversight is moving from optional guidelines to enforceable laws that impose penalties for serious system failures. [#security]]]></description></item>    <item><title>#security — Federated training trains financial models across bank ledgers without moving data.</title><link>https://therawlogs.com/take/4-5</link><guid>https://therawlogs.com/take/4-5</guid><pubDate>Thu, 20 Aug 2026 17:33:41 GMT</pubDate><description><![CDATA[Federated training trains financial models across bank ledgers without moving data.: The age of privacy is gone. Federated learning and federated training modules have emerged. 
Where? In Finance. Yes! They say cross-border bank ledgers are trained without transferring raw customer records. Do we believe them or protest? [#security]]]></description></item></channel></rss>