ThreatFabric's investigation describes Bazuka as a modular Android fraud framework with a loader, a core and 23 modules. Observed components support credential overlays, keylogging, interception of verification messages, remote control and NFC relay activity. A captured home or lock screen can be displayed over the real interface to conceal background actions. The AI element sits in the operator's control panel: researchers found support for local and API-hosted language models, including prompts for victim assessment and permission strategies. That evidence does not establish a fully autonomous fraud campaign or measured AI-driven success rate. The original report also qualifies two claims that can be overstated in secondary coverage. Google Play distribution was an apparent intention, not a confirmed infection route; independent modules surviving removal of the host app were a broader concern about ADB bridging, not a demonstrated Bazuka capability. The practical detection opportunity is the combination of abusive accessibility access, debugging changes, module loading and suspicious network activity.
Bazuka: separate observed mobile attacks from claims about AI autonomy
Related research
More curated notes connected through Agent Security and AI Red Teaming.
Pacing model development in an era of cyber-critical capabilities
OpenAI says preliminary evidence that Astra may meet its Critical cybersecurity threshold led it to pause frontier reinforcement-learning work for two weeks and keep its largest planned run on hold. New safeguards include stronger workload and network isolation, continuous boundary testing, token-level monitoring that escalates suspicious tool activity, and broader alignment checks for deception, reward hacking, and unauthorized access.
OWASP Top 10 for Agentic Applications for 2026
OWASP's community guide organizes agentic-system risk into ten categories, including goal hijacking, tool misuse, identity and privilege abuse, memory poisoning, insecure inter-agent communication, cascading failures, and rogue-agent behavior. It provides a shared taxonomy and mitigation starting point rather than a certification checklist or evidence that a deployed system is secure.
A blueprint for democratic governance of frontier AI
OpenAI proposes a three-part U.S. frontier-AI governance model: harmonize emerging state safety laws into a federal baseline, strengthen CAISI as an evaluation and standards institution, and coordinate a broader resilience program. Proposed controls include severe-risk evaluations, transparency reports, independent audits, safety-incident reporting, model-weight security, whistleblower protection, and periodic technical assessments.