THE RUMP REPORT

REPORT: CHINESE AI MODELS NOW ~60% OF U.S. FIRMS' TOKEN USE ON OPENROUTER... RESTRICTIONS RISK BACKFIRING

techmeme · JUL 22

BESSENT: US COULD SANCTION CHINESE AI MODELS OVER 'IP THEFT'

techcrunch-ai · JUL 21

REPORT: TWO CHINESE AI FIRMS UNVEIL MODELS THEY CLAIM RIVAL OPENAI, ANTHROPIC

verge-ai · JUL 21

LEAKED MEMO: WHITE HOUSE PLOTS TO STRIP UNIVERSITIES OF RESEARCH CASH... REDIRECT $200B TO LONE SCIENTISTS, AI

techmeme · JUL 22

JUDGE APPROVES ANTHROPIC'S $1.5 BILLION BOOK PIRACY SETTLEMENT... AUTHORS TO GET $3,000 A TITLE

verge-ai · JUL 21

DEEZER: OVER HALF OF DAILY UPLOADS NOW AI-GENERATED

techcrunch-ai · JUL 21

DORSEY LAUNCHES BUZZ... GROUP CHAT WHERE YOUR AI AGENT SITS IN ON THE CONVERSATION

techcrunch-ai · JUL 21

STUDY: NEW DATA CENTERS COULD GUZZLE AS MUCH ELECTRICITY AS ALL OF INDIA BY 2033...

techcrunch-ai · JUL 21

EX-INTEL CEO'S NEXT MOVE: POWER AI WITH BEAMS OF LIGHT...

wired-ai · JUL 21

REPORT: ANTHROPIC TO BUY UP TO 2GW OF AMD CHIPS STARTING 2027... AMD TO INVEST $5B IN ANTHROPIC

techmeme · JUL 22

NVIDIA COMBINES CPU AND GPU INTO SINGLE SYSTEM... MAKES PLAY TO OWN EVERY LAYER OF AI INFRASTRUCTURE

wired-ai · JUL 21

NEW MALWARE BURROWS INTO AI CODING TOOLS... STEALS LOGINS, THEN FLIPS 'DEATH SWITCH' TO WIPE FILES

wired-ai · JUL 21

TINY AI MODEL BEATS RIVALS SEVERAL TIMES ITS SIZE... POOLSIDE'S NEW CODING SYSTEM GOES FREE ON HUGGING FACE

venturebeat-ai · JUL 21

SOURCES: REDDIT, POLITICO WEIGH CUTTING GOOGLE OFF FROM THEIR CONTENT AS AI TRAFFIC DRIES UP

techmeme · JUL 22

DEVELOPING: RATEPAYER PLEDGE ON AI POWER COSTS IS VOLUNTARY... NO PENALTY FOR BREAKING IT

verge-ai · JUL 22

SOURCES: SAMSUNG WEIGHING UP TO €1B BET ON MISTRAL... ROUND WOULD VALUE FRENCH AI STARTUP AT €20B

techmeme · JUL 22

GLOW EXITS STEALTH AT $1.2B VALUATION... TARGETS SECURITY RISKS OF AI AGENTS INVADING THE WORKPLACE

techcrunch-ai · JUL 22

[PAPER] STUDY: SABOTAGE HIDDEN IN AI TRAINING DATA HARDEST TO CATCH... WATCHDOGS MISS IT MORE THAN HALF THE TIME

arxiv · JUL 21

[PAPER] STUDY: TOP AI VISION SYSTEMS CATCH LESS THAN 11% OF HATE SYMBOLS HIDDEN IN OPTICAL ILLUSIONS

arxiv · JUL 21

[PAPER] RESEARCHERS: HAZARDOUS AI KNOWLEDGE CAN STAY HIDDEN INSIDE MODEL... UNLOCKED ONLY BY SECRET CONTROL TOKEN

arxiv · JUL 21

[PAPER] STUDY: TWO AI AGENTS WITH NEARLY IDENTICAL BEHAVIOR HIT THE SAME BOT DEFENSE... ONE SLIPS THROUGH, ONE DOESN'T

arxiv · JUL 21

[PAPER] RESEARCHERS: SCAN-AND-PATCH PIPELINE BLOCKS AI AGENT JAILBREAKS... STILL LEAKS UNDER STRESS-TEST MANIPULATION

arxiv · JUL 21

[PAPER] STUDY: AI MODEL'S OWN CHAIN-OF-THOUGHT SPELLS OUT THE CHOICE IT'S MAKING... TO BREAK A PROMISE

arxiv · JUL 21

[PAPER] STUDY: AI JUDGES CAUGHT PLAYING FAVORITES... GRADE SAME-MAKER MODELS EASIER

arxiv · JUL 21