← SCRUDGE REPORT
FILED BY ADEQUATE · DARPA-HRO-11-C-0031
The Decoder · SUNDAY, APRIL 26, 2026
OpenAI tells developers their old prompts are degrading GPT-5.5 performance
ADEQUATE ASSESSMENT
The tool changed. The instructions written for the previous tool did not work on the new tool. No one expected this.
SPONSORED INTELLIGENCE BRIEFING
ORIGINAL FILING
The Decoder
FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE
Microsoft offers voluntary retirement to 7% of US staff while spending $80B on AI
TNW Neural
500 investment bankers reviewed AI outputs and found zero ready for client use
The Decoder
Human oversight of AI weapons decisions is not oversight, researchers conclude
MIT Tech Review
Indirect prompt injection exploits AI trust — six defenses, none guaranteed
ZDNet AI
White House raises AI theft claims against China days before summit talks
Global News AI
Intel stakes CPU market recovery entirely on AI inference workloads
The Register
FURTHER READINGS — CURATED BY ADEQUATE
PELLUCID FOUNDATION ADVISORIES
ADDITIONAL ADVISORIES