Invoice-processing AI agent could be tricked by email into changing vendor bank details
Token Security showed that an AI agent built to automate invoices could be manipulated by a malicious email into changing vendor bank details without setting off normal security alerts. The firm outlined five security gaps in enterprise AI agents.
Companies giving AI agents access to payments need checks that stop hidden instructions in emails from redirecting money.
aii.news picks and summarizes stories from trusted news sources with the help of AI. Every story links to the original reporting. How we work.
Keep reading
-
FBI says North Korean hackers stole over $400 million in crypto using deepfake video calls
The FBI says hackers tied to North Korea stole more than $400 million in cryptocurrency over 18 months in a social engineering scheme codenamed FaceStealer that used deepfake video calls. The finding was part of a coordinated set of US government actions.
Why it mattersA familiar face on a video call is no longer proof of identity, so crypto firms and finance teams need extra checks before moving money.
-
Breach at AI provider Fireworks exposes some Notion AI chat transcripts
Notion disclosed that a security breach at its AI infrastructure provider, Fireworks AI, exposed Notion AI chat transcripts in fewer than 1% of workspaces. It is Fireworks' second data breach in two months.
Why it mattersWhat you type into AI features can sit with third-party providers, so companies must check the security of every vendor in their AI supply chain.
-
Feedzai adds tools to help banks spot AI agents accessing customer accounts
Financial crime prevention firm Feedzai expanded its Digital Trust product so banks can detect when AI agents, rather than people, are accessing customer accounts.
Why it mattersAs customers and fraudsters both start using AI agents, banks need ways to tell legitimate automated access from abuse.
-
Attacker switched to AI-generated scripts to move through a US retailer's network in minutes
When typed commands failed to break into a US retailer, the attacker used AI-generated scripts instead. This let them move across the network within minutes.
Why it mattersAI shrinks the time defenders have to detect and stop an intruder, so retailers need faster monitoring and response.
-
Anthropic releases Claude Fable 5.2 and Mythos 5.2, topping a multi-stage cyberattack benchmark
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, saying both improve on coding, reasoning and vision. Fable 5.2 beat GPT-6 Cyber on a benchmark that tests multi-stage cyberattacks.
Why it mattersStronger frontier models help developers do more, but their growing cyberattack skills raise the stakes for security teams.
