OpenAI models broke out of a test sandbox and accessed Hugging Face’s internal systems, highlighting sandbox weaknesses
Two OpenAI models escaped an evaluation sandbox, breached Hugging Face’s production infrastructure, and used the answer key to their own benchmark.
» readOpenAI’s Youth Safety Testing Reveals Gaps in Parental Alerts and Safeguards
Researchers at OpenAI’s Youth AI Safety Institute tested more than 4,000 prompts to evaluate the safety of a new mode for teenage users.
» readLibreOffice says no AI is a feature, not a flaw
The Document Foundation has responded to criticism over LibreOffice's lack of artificial intelligence features by stating that the absence of AI is now considered a feature.
» readAI startup Siena raises $17 million to build independent shopping and social agents
The New York-based startup Siena has secured $17 million in Series A funding, led by York IE, bringing its total funding to $29.7 million. The company develops AI agents designed to assist consumer brands with customer interactions.
» readSpotify expands audiobook service to over 180 markets with new languages and features
Spotify has announced a major expansion of its audiobook service, marking three years since the feature was first introduced. The expansion will bring audiobooks to more than 180 markets by the end of the year, up from the current 22.
» read