Data leak stories
Australian businesses risk data leaks and governance gaps as staff adopt AI tools faster than employers can set rules and training.
Prompt-injection attacks and data leaks are the main risks as businesses connect Copilot and Gemini to email, files and internal tools.
Companies can now trace AI prompts to individual staff, helping curb surprise bills and reduce the risk of sensitive data leaks.
Boards are under pressure to curb staff use of ChatGPT and similar tools, as pasted data could expose customer records and source code.
Security teams can now monitor Claude-based agents for prompt injection, data leaks and rogue tool access as autonomous AI use spreads.
Centralised oversight for AI traffic should help firms cut prompt-injection and data-leakage risks as models shift into production.
Security checks are delaying most AI projects for weeks or months, with many firms deploying watered-down systems to get them live.
Downstream AI data leaks have more than doubled in a year, with connected services now returning unauthorised information to staff and agents.
Organisations adopting Microsoft's top security tier are being offered a single bundle to cover data protection, identity controls and AI agent oversight.
Readers can now inspect breach reports by source, verification and context, as the service avoids inflating record totals into a unique-victim count.
Security teams gain tighter endpoint oversight of shadow AI and sensitive data, as Fortinet folds new controls into FortiEndpoint from the third quarter.
Only a small share of alerts proved urgent, but critical vulnerability exposures more than doubled as phishing also surged in the report.
Users of Dify's cloud service could have had private chats and files exposed after Zafran Security disclosed four flaws in the AI platform.
ESET says the gang's operator-backed toolkit could help affiliates bypass defences faster, widening the threat to businesses worldwide.
Ransomware losses worsened in May as attacks climbed 48% year on year, despite a 7% drop in overall cyber incidents.
Many firms are exposing sensitive data as shadow AI and weak controls leave them open to breaches, hallucinations and unauthorised access.
New silicon-level controls aim to curb unauthorised agent access and data exposure in enterprise AI storage, while keeping traffic fast.
A recent government breach has exposed a wider gap in cyber awareness, with stolen personal data sold online for as little as GBP £5.
A faulty token check let low-privilege users view other applicants' identities, payment details and Social Security numbers in a financial onboarding app.
Ransomware attacks spread through vendor systems and AI-aided tactics, leaving 7,551 publicly disclosed victims in the latest 12-month period.