If you enjoy this newsletter, please become our paid subscriber to help this keep going.
Highlights
McDonald’s AI get hacked.Scammer impersonate US secretary.SB53: Transparency policy.EU: General Purpose AI code of practice.New powerful AI models are coming.
Special!
Deep Dive
McDonald’s AI get hacked Wired
AI hiring bot password leak.
Exposed 64 million records.
Applicant’s data and tests.
Hackers tried “123456”.
9 July: Found by security researchers.
Scammer impersonate US secretary APNews
Register an account on Signal.
Used email “mario.rubio@state.gov”.
Used AI tool for text and voicemails.
3 July: Contacted US high ranking officials.
Unclear who was the behind.
SB53: Transparency policy Senate.ca.gov
Free and low costs access to GPUs
Protect whistleblowers at AI labs.
To publish safety and security protocols.
Report incidents within 15 days.
“Trurst, but verify” approach.
EU: General-Purpose AI Code of Practice Europa.eu
13 independent experts, 1000 stakeholders.
Enter to application 2nd August 2025.
Three chapters.
Transparency, Copyright, Safety and Security.
Still need EU endorsement to proceed.
New powerful AI models are coming
AI help Elon Musk answer question. Grok4
Upcoming launch of the open weight mode. OpenAI.
Comet: New browser with AI. Perplexity
Notice: Y2 GenAI Safety and Security is on GumRoad and Amazon with paperback.
Share this post