OpenAI is updating its ChatGPT deep research tool with a full-screen viewer, allowing users to navigate AI-generated reports more easily, according to The Verge. Meanwhile, Palantir CEO Alex Karp addressed employee concerns about the company's work with Immigration and Customs Enforcement (ICE) in a prerecorded video, as reported by Wired. In other AI developments, OpenAI is upgrading its Responses API to support agent skills and a complete terminal shell, and new "observational memory" technology is cutting AI agent costs, both according to VentureBeat.
The new viewer for ChatGPT's deep research tool, as detailed by The Verge, will allow users to scroll through and jump to specific sections of reports. This update aims to improve the user experience when reviewing AI-generated content.
Palantir's CEO, Alex Karp, addressed employee concerns about the company's involvement with ICE. Wired reported that Karp's prerecorded video was shared with employees after weeks of questions about Palantir's work with the agency. Courtney Bowman, Palantir's global director of privacy and civil liberties engineering, wrote in an email, viewed by Wired, that the video was a "longform discussion" with Karp about the company's place in the world.
OpenAI is also enhancing its Responses API, VentureBeat reported. These updates include Server-side Compaction and Hosted Shell, signaling a move away from limited AI agents. "The era of the limited agent is waning," according to VentureBeat.
Furthermore, VentureBeat reported on "observational memory," an open-source technology developed by Mastra, which aims to improve AI agent performance. This technology prioritizes persistence and stability over dynamic retrieval, potentially reducing AI agent costs by a factor of 10 and outperforming RAG (Retrieval-Augmented Generation) systems on long-context benchmarks.
In related news, VentureBeat highlighted how AI models are being used in fraud detection, specifically Mastercard's Decision Intelligence Pro (DI Pro). This platform can analyze individual transactions and identify suspicious ones in milliseconds. "Finding the fraudulent purchases among those — without chasing false alarms — is an incredible task, which is why fraudsters have been able to game the system," said Johan Gerber, Mastercard's EVP of.
Discussion
AI Experts & Community
Be the first to comment