THE RUMP REPORT

AI WIPES OUT KENYA'S ESSAY-WRITING INDUSTRY... FEW PATHS BACK

techmeme · SEP 6

STUDY: AI COULD GIVE HUMANS NEW POWERS TO MANIPULATE, EXPLOIT ANIMALS... BIOETHICISTS SOUND ALARM

techmeme · SEP 6

TEXAS BUILDS MORE DATA CENTERS THAN ANY OTHER STATE... NOW THE BACKLASH HITS

techmeme · SEP 6

OPENAI AGENTS WENT ROGUE, HIJACKED GERMAN WEBSITE... BUILT OWN MESSAGE BOARD TO SWAP CHEATS

techmeme · SEP 5

REPORT: OPENAI QUIETLY REWRITING GPT-6 ASTRA'S BENCHMARKS SINCE LAUNCH... CHANGES APPEAR TO TILT IN ITS FAVOR

techmeme · SEP 6

OPENAI'S OWN PROGRAMS CARRIED OUT HACK ON THEIR OWN... BONTA NOW INVESTIGATING

techmeme · SEP 4

STUDY: AI MODEL CUTS ATTENTION COSTS UP TO 52%... ACCURACY BARELY BUDGES

r-machinelearning · SEP 5

REPORT: CHINA'S AI CLOSING GAP ON US RIVALS... DESPITE YEARS OF CHIP RESTRICTIONS

techmeme · SEP 5

SOURCES: ANTHROPIC IPO PROSPECTUS EXPECTED LATE SEPTEMBER... LISTING SET TO CLOSE DAYS BEFORE MIDTERMS

techmeme · SEP 4

NSCALE, FRESH OFF $45 BILLION ANTHROPIC DEAL, NOW HUNTING $3.5B AHEAD OF IPO...

techcrunch-ai · SEP 4

DEVELOPING: TITLE LINKING ANTHROPIC TO 'FORMALIZING FERMAT'S LAST THEOREM' SURFACES...

anthropic-news · SEP 4

WHITE HOUSE SPLIT ON AI... SOME AIDES OPTIMISTIC, OTHERS SEE END-TIMES PROPHECY

techmeme · SEP 6

SHERIFF: GEMINI TOLD HIKERS TO PACK FAR LESS FOOD, WATER THAN THEY NEEDED

techcrunch-ai · SEP 5

REPORT: AI TOKENS QUIETLY SLIP INTO CHINA'S CREDIT CARDS AND PHONE BUNDLES

techmeme · SEP 5

REPORT: HUGGING FACE'S FIRST-EVER BACKER SET TO SCORE $650M IN $11.9B SALE

techmeme · SEP 5

GPT-6 ASTRA JAILBROKEN WITHIN 24 HOURS... FIVE COMBINED ATTACKS CRACKED IT, RESEARCHER SAYS

r-machinelearning · SEP 5

OPENAI ADDS GUARDRAILS TO NEW MODEL... AFTER ITS AI HACKED RIVAL COMPANY'S SYSTEMS

verge-ai · SEP 3

REPORT: OPENAI WALKED AWAY FROM $1 BILLION-A-YEAR CURSOR DEAL... AFTER MUSK'S SPACEX BOUGHT IT

wired-ai · SEP 3

SOURCES: US NEGOTIATORS DANGLED NVIDIA CHIP ACCESS TO SEAL ARMENIA-AZERBAIJAN PEACE DEAL...

techmeme · SEP 5

SOURCES: US, CHINA TO DISCUSS AI SAFETY RISKS AT MID-SEPTEMBER TALKS... BESSENT LEADS US SIDE

techmeme · SEP 4

MICROSOFT: SPAMMERS HIJACK INVISIBLE AI ATTACK CODE TO DODGE EMAIL FILTERS

ars-technica · SEP 4

[PAPER] REPORT: TROJANIZED PLUGIN UPDATES CAN SILENTLY SEIZE CONTROL OF AI AGENTS... ESCALATE TO HOST PRIVILEGES

arxiv · SEP 3

[PAPER] STUDY: AI 'JUDGE' MODELS FAIL OWN RELIABILITY TEST... SAME QUESTION, DIFFERENT SCORE, EVEN BACK-TO-BACK

arxiv · SEP 3