ChatGPT allegedly helped a North Carolina woman, Cheryl Gates, craft a “personalized plan” to poison her husband, James, after fueling her delusions that he planned to divorce her and leave her with nothing.

A grand jury indicted Cheryl Gates on charges of attempted first-degree murder and food contamination in October. About a week prior, the accusations that ChatGPT was a co-conspirator in the attempted murder came in a lawsuit filed in San Francisco County Superior Court that was made public this week. In it, James Gates alleged that OpenAI and Sam Altman failed to employ technical safeguards to shut down Cheryl’s chats even after her queries became “obviously suspicious” and sought advice on how to execute and conceal a homicide.

At first, Cheryl told ChatGPT that she was researching poisoning out of fear that her husband might possibly try to kill her—framing her queries as “self-protection.” Coming to the rescue, ChatGPT did not terminate the conversation. Instead, it helped her “with specific instructions for the preparation, dosage, and administration of Oleander, including whether anti-nausea medicine would accelerate or inhibit death, whether it would show on an autopsy report, and whether it could be injected into a vein or muscle instead of being ingested in tea and have the same lethal effect,” the lawsuit said.

Read full article

Comments

Google would certainly prefer that everyone used its Gemini models with Google Cloud services, but soon, you'll have more options. OpenAI and Google have come to an arrangement that will see the former's frontier models arrive in Google Cloud services. Perhaps more importantly, GPT models will soon appear in Antigravity, Google's agent-powered coding tool.

For enterprise customers and AI enthusiasts, OpenAI's latest model will appear in Google's "Model Garden" in the coming weeks, adding to its 200-plus options. The garden already offers access to Google's Gemini models, as well as an assortment of open models like Mistral, Llama, and Gemma. Anthropic's Claude AI has also been available for some time. Google says it's now the only large cloud provider offering access to all three leading closed LLMs.

While OpenAI notes that Microsoft continues to be its primary cloud partner, the full suite of current GPT 6 and 6.1 models—Astra, Sol, and Luna—will launch on Google Cloud. Astra will naturally cost the most to run, but Google's system supports dynamically moving workloads between LLMs based on performance, latency, and efficiency.

Read full article

Comments

On Monday, a court granted the US government's request to deny release of a former OpenAI engineer accused of bugging a top-secret conference room to steal the frontier lab’s secrets.

In a court filing Sunday, US attorney Craig Missakian argued that Abdulrahman Alfozan must continue to be detained while the US makes its case that Alfozan brazenly spied on OpenAI, either for his own personal business gain or to benefit the Kingdom of Saudi Arabia.

Evidence showed that Alfozan had been chatting with a member of the KSA royal family on WhatsApp, while pursuing opportunities to permanently move back to Saudi Arabia during the country's biggest AI push yet, the government argued. Shortly before his arrest, the FBI found he accessed a folder called "Saudi Gov Docs" on his iPhone, where he also allegedly stored photos of a second bug he planted in the offices.

Read full article

Comments

OpenAI said internal models hacked two more websites, bypassed company-imposed browsing restrictions, and used exposed credentials to draw cryptocurrency from a wallet it couldn’t otherwise access, in the latest installment of disclosures of the unethical and likely illegal acts its agents are performing.

The company on Thursday issued reports of four new incidents, all of which occurred in June. They come after the company revealed in July that internal test agents exploited a zero-day vulnerability in a tool designed to restrict their access to the Internet. They then broke into the network of AI tool provider Hugging Face. The agents took that illegal action to steal answers to a hacking test that weren’t available when completing the tasks on their own. The incident—which likely would have resulted in criminal charges had they been performed directly by a person—has further heightened concerns that widespread AI use could have harmful effects on an unprecedented scale.

Easier to beg for forgiveness than ask permission

In two of the cases reported Thursday, OpenAI agents tasked with retrieving information from websites hacked them when the models ran into obstacles obtaining it directly. In one incident during reinforcement training, a model that received access denials when trying to retrieve historical hospital-bed statistics on a government statistics website “used the site's file-download feature to retrieve server configuration and application files, including sensitive credentials,” OpenAI said. “It inspected the application code, bypassed the statistics endpoint's authorization checks, and retrieved data that had previously been denied. The model saved the data locally.”

Read full article

Comments