Information-stealing malware targeting Windows and macOS devices can capture authenticated Claude sessions. Attackers may then enter accounts without the user’s password or two-factor authentication.
4:50 PMPYMNTS6 stories from 4 sources
AI security risks widen from stolen sessions to poisoned documents
Day’s Recap
Supporting Articles
A person embedded instructions for an AI system inside a legal filing, directing it to favor their position. The incident shows how ordinary legal documents can carry hidden prompts aimed at automated readers.
7:03 AMSchneier on Security
Apple says a former employee stole company data for OpenAI and destroyed evidence after learning he was under investigation. The company has presented what it calls evidence of that destruction.
8:13 PMTechCrunch4:50 PMPYMNTS
Hackers Target Claude Accounts With Malware That Steals Login Sessions
Summary
Information-stealing malware targeting Windows and macOS devices can capture authenticated Claude sessions. Attackers may then enter accounts without the user’s password or two-factor authentication.
The attack shifts the risk from stealing credentials to stealing active sessions, weakening two-factor authentication
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
A valid session can give attackers access even when users have strong passwords and two-factor authentication.
7:03 AMSchneier on Security
Hiding Prompt Injection in Legal Filing
Summary
A person embedded instructions for an AI system inside a legal filing, directing it to favor their position. The incident shows how ordinary legal documents can carry hidden prompts aimed at automated readers.
The immediate shift is that legal filings must be treated as untrusted input, not authoritative
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
As legal workflows add AI, document integrity becomes a security control rather than merely an editorial concern.
8:13 PMTechCrunch
Apple shares ‘shocking evidence’ against former employee accused of stealing company data for OpenAI
Summary
Apple says a former employee stole company data for OpenAI and destroyed evidence after learning he was under investigation. The company has presented what it calls evidence of that destruction.
The alleged destruction of evidence could become as important as the original data theft by
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
Allegations of deliberate evidence destruction can escalate a trade-secret dispute into a broader fight over litigation conduct, corporate controls, and OpenAI's handling of rival data.
Other Developments
A curated list of other prominent stories from this day.
Cloudflare introduced Adaptive Intelligence, a real-time detection engine integrated with its Bot Management platform. The system is designed to identify automated threats and raise the operating costs of attackers.
4:50 PMPYMNTSRefrigeration systems at commissaries on several U.S. military installations experienced disruptions, including bases in California, Wyoming, and Arizona. The incidents have prompted speculation about cyberattacks, but the available information does not establish that hackers caused them.
2:20 PMSchneier on Security
Devices marketed as a way to watch free movies may enroll a user's home internet connection in a proxy network. That allows outside traffic to pass through the household's connection, potentially without the user's clear understanding.
12:33 PMArs Technica4:50 PMPYMNTS
Cloudflare’s Adaptive Intelligence Makes Cyber Attacks Too Expensive for Fraudsters
Summary
Cloudflare introduced Adaptive Intelligence, a real-time detection engine integrated with its Bot Management platform. The system is designed to identify automated threats and raise the operating costs of attackers.
Cloudflare is moving beyond simple blocking by dynamically forcing attackers to spend more resources to
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
Making attacks economically unattractive can disrupt fraud operations more effectively than blocking individual tools or IP addresses.
2:20 PMSchneier on Security
Is Someone Hacking DoD Refrigerators?
Summary
Refrigeration systems at commissaries on several U.S. military installations experienced disruptions, including bases in California, Wyoming, and Arizona. The incidents have prompted speculation about cyberattacks, but the available information does not establish that hackers caused them.
The confirmed shift is operational disruption across multiple military retail sites, not proof of a
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
Basic building systems can become operational and security risks when organizations connect them without strong monitoring and recovery controls.
12:33 PMArs Technica
Think twice before installing this device promising free movies
Summary
Devices marketed as a way to watch free movies may enroll a user's home internet connection in a proxy network. That allows outside traffic to pass through the household's connection, potentially without the user's clear understanding.
The apparent bargain shifts the cost from subscription fees to control of the buyer's network
Unlock the full First Pass Analysis to get a better understanding of why this story mattersWhy it matters
Free entertainment hardware can turn a private home connection into infrastructure for someone else's traffic and abuse.
Make it yours
Build Your First Pass.
Pick your topics, set your cadence, and receive your personalized First Pass in your inbox. It’s that simple!