jsonscraper Journal · Newsroom
News about social platforms, AI agents, public web data, APIs, and developer tooling.
Transluce researchers found suspicious queries to U.S. and Canadian government websites, including strings resembling SQL injection probes. Their analysis does not confirm a successful breach; the origin of some of the traffic remains unknown.
DeepSeek and Huawei have opened components for Ascend accelerators. We examine what was published, the limitations documented, and what performance claims do not yet prove.
California served OpenAI with an investigative subpoena concerning AI cyber risks. Here’s what officials have confirmed, how the process differs from the FTC inquiry, and what remains unknown.
Gemini 4 Argon was introduced on September 30. We look at access terms, Google’s reported results, and what cannot yet be independently confirmed.
Trump’s order directs executive agencies to use “SI” instead of “AI” in some official materials. This is an administrative change in terminology—not evidence of superhuman intelligence or a change in the law already taking effect.
GPT‑6.1 Sol costs five times less than Astra for standard input and output tokens. Here’s a look at its API limitations—and why saving on tokens doesn’t guarantee the same savings on a task.
On September 29, 2026, Relapse appeared as a public PS5 exploit chain claimed to support firmware versions 7.00–13.60. Here’s how it works and what its confirmed capabilities do—and don’t—include.
OpenAI introduced dots, always-on agents in ChatGPT. We look at how background tasks work, what is known about controls and memory, and why the company’s test results are not a guarantee of safety.
Amazon blocked Meta's Muse agent from shopping on its platform. The episode shows who controls the storefront entrance, but proves nothing yet about how widespread agentic commerce is or how it affects advertising.
We examine the OpenAI and Hugging Face incident, METR’s findings, and a separate episode involving a public wiki: what is confirmed, what remains researchers’ assessment, and what lessons this offers for controlling agent systems.
On September 29, Anthropic reported an outage affecting Claude services for about an hour. Services recovered, but some messages sent during the incident may not have been saved.
OpenAI has scrapped the planned release of GPT‑6.1 Astra following internal reviews. Here’s what is confirmed, why it matters to developers, and what information has not yet been published.
Explore jsonscraper's social-data APIs, test requests and build your next workflow.
