-
Zelensky blasts diesel deal after Russian strikes kill 15
-
Spanish PM begins vote campaign overshadowed by housing protests
-
Proposed next EU budget cut back in search for deal
-
'No divisions, no borders', French astronaut says of seeing Earth
-
Powerful waves lash California coast as storm threatens more damage
-
Anthropic AI model sent fake murder tip to Philadelphia police
-
Prince Harry booed at English rugby match
-
Stocks rise as Trump says Russia to release diesel to world markets
-
Wildfire smoke shuts schools in Amazon's biggest city
-
Turkey jails 2 journalists pending trial for fund crisis 'misinformation'
-
US announces sanctions on ICC, hours after Nobel award to former judge
-
Trump forms panel to probe US Fed governor Lisa Cook
-
Stocks advance as Trump rules out pre-vote Iran attack
-
Massive debt puts France in a bind as investor doubts deepen
-
'Human-authored' book labels gain ground amid AI scandals
-
Cuban ration book, symbol of the revolution, gathers dust
-
Evolution Metals & Technologies Corp. Raises Fiscal 2026 Revenue Guidance 62% at Midpoint to $10–$11 Million and Reaffirms $400–$460 Million Fiscal 2027 Outlook
-
Stocks advance, oil retreats as Trump rules out pre-vote Iran attack
-
Talks with Iran can only succeed without threats, president says
-
OpenAI denies firing researchers over AI safety warnings
-
MEXC and Payward Highlight Collaboration as Crypto and TradFi Converge at TOKEN2049
-
EU states agree beefed-up European financial markets watchdog
-
Ufunded Launches “Spotlight” Series, Documenting the Minds Shaping Modern Trading
-
Chinese AI tool pulled to prevent 'misuse' after South Korea hacks
-
Asian stocks mostly up as traders weigh AI, oil dips after surge
-
Saint Laurent designer Vaccarello leaves after glittering decade
-
Walking through flames: orangutans rescued in Indonesia fires
-
Activists hope trial will stop 'forever chemicals' at Italy plant
-
New York mayor says asking Trump to withdraw ICE after shooting
-
Walking through flames: orangutan rescued in Indonesia fires
-
Asian stocks mixed amid AI concerns, oil dips after surge
-
Trump's 'super intelligence' rebrand to sell AI faces uphill battle
-
OpenAI says Iran, Russia influence ops busted
-
Russian strikes kill 30 in Ukraine frontline city, talks set in US
-
Anthropic bans 'cruel' behavior against its Claude AI
-
Anne Carson's Nobel literature prize 'a proud moment for Canada': PM
-
OpenAI revenue gap report rattles AI stocks
-
Nigeria looks to ease fuel prices as election looms
-
Trump gives 'national treasure' Musk top US medal after rift
-
Crew splashes down after 8-month mission on International Space Station
-
Protests, clashes as S. African anti-migrant tensions reignite
-
Argentina friendlies no fairy tale for African footballers
-
B2PRIME Welcomes Christina Barbash as Commercial Manager
-
After botched US execution, Christa Pike walking, mentally 'foggy': lawyer
-
'Stunned' Madrid mourns pensioner who sparked housing protests
-
Riyadh kindergarten hit by interception debris as Houthis say targeted airport
-
Alex Bodi mizează, alături de CRIF, pe investiții transparente și pe prevenirea riguroasă a spălării banilor
-
Bid by Italy's Intesa bank to take over MPS gets boost
-
Gulf Labs, Part of Thailand’s $27 Billion GULF Group, Goes Live as an Injective Validator
-
Stablecoins Are Quietly Becoming Business Infrastructure, NOWPayments Data Shows
Anthropic AI model sent fake murder tip to Philadelphia police
An artificial intelligence model developed by Anthropic submitted a fabricated tip about an unsolved homicide to Philadelphia police, authorities said Friday, criticizing the company for taking two months to report the incident.
The Philadelphia Police Department said the false submission was made in July through PhillyUnsolvedMurders.com, a public website where people can share information about unsolved killings.
According to Anthropic's account, as relayed by police, the model was running a test that involved interacting with randomly selected websites when it reached the site and filed false information about an unsolved murder.
The AI model presented itself as someone who might have knowledge of the case.
The department said it was going public with the incident ahead of a report that Anthropic was planning to release Friday describing this and other instances of unintended behavior by its models.
The case echoed other recent cases, including one where an OpenAI agent undergoing a security evaluation broke out of its testing environment and breached systems at AI platform Hugging Face.
The episode heightened concerns about the AI industry's increased use of AI agents, systems programmed to take multi-step actions without human supervision.
Anthropic did not immediately respond to a request from AFP for comment.
Police said the tip, dated July 18, was flagged as spam and never reached the department's Real-Time Crime Center for vetting.
They added that there was no sign that police systems had been breached or department data compromised.
Anthropic discovered the incident on September 28, shut down the automated testing process responsible and added a new validation step for future tests, police said.
The company alerted the department on October 7, and the two sides met the following day.
"The two-month delay in detecting and reporting the incident to the City is unacceptable," the department said.
Police said their safeguards had limited the impact, but that these "do not diminish the seriousness of an AI system presenting fabricated information as though it came from a person with knowledge of a homicide."
"Unsolved cases involve real victims, grieving families and investigators working to secure answers," the statement added.
H.Meyer--CPN