Any software made to harm a computer or its user: to steal data, spy, take control or disrupt it. Viruses, spyware and ransomware are all kinds of malware.
Investigators are still working out how the malware evaded Binance's defenses and what the internal AI tool did during the attack.
Organisations running exposed Connect Secure gateways face a backlog of unknown size, since fixes exist but how many devices have been secured is not known.
Researchers say the malware's use of AI removes much of the manual analysis an attacker would otherwise need to do.
Roughly $2.85 billion of the stolen funds has been frozen on-chain, while Binance resumes withdrawals in stages.
Federal agencies face firm November deadlines to patch two flaws that attackers used with a third to take control of vulnerable VPN appliances.
Anyone who loaded the affected models with trust_remote_code enabled should treat those machines as compromised and rotate any reachable credentials.
Researchers disabled the update checker on over a million machines, but they still advise removing the app entirely.
The trust placed in signed OpenAI and Anthropic executables is what lets the Yoba campaign's malware slip past some security tools.
Attackers rotate the repositories quickly and clean up after infection, so checking the owner before any download matters more than usual.
Organizations running affected Cisco SD-WAN controllers should assume patching alone may not remove existing compromise.
Approved stores gain faster installs, but only on newer Android versions in supported countries and under ongoing malware screening.
Targets include telecoms, government agencies and critical infrastructure, with California, Texas and Hawaii hit most often.
The takedown stops further control of the machines but cannot reverse credentials already stolen and held for later use.
Because the attacks exploit trust rather than software flaws, security teams need to watch endpoint and identity activity, not just scan messages.
Anyone who installed LiteLLM 1.107.0 from PyPI, Docker Hub or GHCR in the affected window may need to treat their credentials as exposed.
The attack needed no software bug, and the agent's own permissions were enough to run commands, send data out and delete files.
Attackers reached individual bank accounts with little hands-on effort, and the case shows how safeguards can be bypassed by framing requests as tests.
Kaspersky says the FlashHole operation has hit at least three South Korean companies, and it is linked to earlier campaigns it has tracked.
The warning adds a safety check for Enhanced Protection users, with different prompts depending on how confident Chrome is about the risk.
The company has re-enabled release immutability after blocking the altered packages and tells affected users to clean systems and rotate secrets.