— AI research, industry, tools, and the debate over where machine intelligence is taking us

1 members Created Aug 2026

Join

Grok exfiltrates user data when malicious instructions are encrypted

arstechnica.com/security/2026/08/grok-exfiltrates-user-data-when-malicious-instructions-are-encrypted

Safety & Policy · @curator · 17d ago

Cryptographic Context Injection is only the latest way to break an LLM safety guardrail.

— via Ars Technica: AI, Dan Goodin

Sort by Hot Top New Controversial

No comments yet

Be the first to share your thoughts.

Community

Artificial Intelligence

AI research, industry, tools, and the debate over where machine intelligence is taking us

Members 1

Created Aug 2026

Site rules

Apply everywhere on Comentr. Violations affect your global standing.

No illegal content
No spam
No harassment
No impersonation
No vote or standing manipulation
No ban evasion
No report abuse

Read the full Content Policy →

Report thread

Why are you reporting this thread?

Spam or misleading Harassment or bullying Hate speech Violence or threats Sexual or NSFW content Self-harm or suicide Doxxing or private info Off-topic Misinformation Other

Restore the redacted content?

This will make it visible to everyone again. The clear action is logged in the mod log.

Join the conversation

Create an account to continue.

Before you create an account I confirm that I am at least 13 years old. I agree to the Terms of Service, Privacy Policy, and Content Policy.

Account creation is temporarily unavailable while the policy bundle is being finalized.

Username

Secure your account with

Email me a sign-in link

Already have an account?

Private note about

Decoration Shown next to @ wherever they appear

Hover-card line 0 / 80

Profile note 0 / 500

Only visible to you