New ChatGPT Lockdown Mode Limits Tools That Could Enable Data Exfiltration

2026-06-07T01:24:08Zfd508fdfbcc8a9e0f87b6b8e841a8825e88cedcb622d3500f142398d9400dec8
active-exploitationai-discoverycloud-compromisedosmalwaremobile-spywarepatchesphishingpoisoned-packagesproxy-networkrcesupply-chainvulnerabilityweb-shellworm

What happened

Collection of security headlines covering a wave of high-impact vulnerabilities, active exploitations, and large-scale campaigns. Notable items: OpenAI rolled out a ChatGPT Lockdown Mode to mitigate prompt-injection/data exfiltration risks; Bright Data SDK is turning consumer devices (including smart TVs) into residential proxy exit nodes; CISA added a SolarWinds Serv-U DoS (CVE-2026-28318) to its KEV list; an autonomous AI agent discovered 21 zero-days in FFmpeg while Chrome 149 fixed a record number of bugs; multiple supply-chain incidents (Miasma and IronWorm) hit GitHub and npm; Cisco SD‑W

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
the_hacker_news
Record identifier
fd508fdfbcc8a9e0f87b6b8e841a8825e88cedcb622d3500f142398d9400dec8
Enrichment time
2026-06-07T01:24:08Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.