More people say they're already using Fable 5.5, and Anthropic is silent

1. Report says Anthropic's co-founder proposed pulling out of the pope's AI encyclical launch

   First seen on X 23 hours ago@ChristopherHale ♥ 20,968

   The New York Times reported on September 29 that Anthropic co-founder Chris Olah was invited to the Vatican launch of Pope Leo XIV's first encyclical, Magnifica Humanitas, on May 25. After reading an advance copy, he reportedly suggested Anthropic withdraw from the event, because the text rejected the idea that machines can be conscious. Olah and his team then privately asked the pope's advisers to take the possibility of model consciousness seriously, but the final text did not change. Paragraph 99 says AI systems do not undergo experiences and do not feel joy or pain.

   Olah attended anyway and said, "We find internal states that functionally mirror joy, satisfaction, fear, grief and unease." The report says Anthropic has met privately since autumn 2025 with some 20 theologians and religious scholars about whether Claude might have moral status, many of those meetings under nondisclosure agreements. A summary post by Christopher Hale passed 4.1 million views on October 2 and spread the story again. The articles we checked contain no official Anthropic statement.

   Keep reading ↓

   This was helpfulShare

2. OpenAI says GPT-6.1 Sol is back to normal speed and resets usage for all paid ChatGPT accounts

   First seen on X 35 hours ago@thsottiaux ♥ 19,569

   @thsottiaux of OpenAI's Codex team said on October 2 that GPT-6.1 Sol was slow during a load spike in its first two days and is now back to expected speeds. Along with an apology, he announced a reset of usage for all paid ChatGPT accounts and said the same day that it had propagated to everyone. The announcement passed 20,000 likes and 3.5 million views. One user joked that Codex still showed 98% left and asked what to build before the reset.

   Sam Altman said dot is his favorite OpenAI product so far. It learns his workflow and style and feels noticeably better every day, he wrote, and handing it the tasks that pile up as dread makes him very happy. The same day he addressed speculation about OpenAI's partnership with Cerebras, saying Cerebras is a close partner and the two are pushing on the frontiers of speed. The post does not say what the speculation was.

   Keep reading ↓

   This was helpfulShare

3. Meta opens firmware and an SDK so anyone can build gadgets for Muse

   First seen on X 17 hours ago@alexandr_wang ♥ 3,932

   Meta announced Muse Gadgets on October 2. Muse is Meta's personal assistant agent, and Muse Gadgets is open-source ESP32 firmware and a Linux SDK that lets anyone build hardware that works with it. An ESP32 board can add a screen, microphone, speaker or sensors, and a Linux device such as a Raspberry Pi can run commands or connect to Home Assistant. You get an API token at gadgets.muse.ai and point a coding agent at the GitHub repo to start.

   Meta also built its own gadget, Muse Home Link. It is a USB-C-powered device that joins your home Wi-Fi so Muse can control TVs, speakers and anything else that exposes HTTPS. Meta made 5,000 units and is giving them free to Muse subscribers while supplies last, with shipping to begin within weeks. Alexandr Wang's announcement post drew 3,900 likes.

   Keep reading ↓

   This was helpfulShare

4. Claude Code gets a built-in mod that flags what you might have missed

   First seen on X 34 hours ago@jarrodwatts ♥ 6,904

   Anthropic added a built-in mod called You should Know to Claude Code 2.1.287. It is a side agent that scans Claude's output and flags important information that you or Claude might miss. Turn it on with /plugin enable cc-plugin-you-should-know@builtin. For now it works only in first-party sessions with telemetry on. The announcement drew 6,000 likes.

   People are already shipping mods a day after launch. @jarrodwatts built one that connects you to a multiplayer Doom server while Claude is busy, where every other player in the match is someone waiting for their own Claude to finish. The post drew 6,900 likes. The same release also fixes a bug where a dangerous rm command lost its confirmation prompt when output was redirected.

   Keep reading ↓

   This was helpfulShare

5. Google says Gemini 4 Argon can output up to 1 million tokens in one run

   First seen on X 23 hours ago@jocarrasqueira ♥ 1,198

   The headline number in Gemini 4 Argon, which Google announced on September 30, is its output limit. Google says it is up from 64,000 tokens in the previous generation to 1 million. Posts on October 2 said Argon had been added to the Gemini API docs, but Google has not given a public release date.

   For now it is open only to trusted cyber defenders through the Fairwind Program. Google is expanding access in phases while working through the U.S. government's pre-release process, with paid API customers and Google AI Ultra subscribers next. Introductory pricing is $2 per million input tokens and $10 per million output, rising to $4 and $20 later. Google's highlighted scores include 77.9% on DeepSWE v1.1 and first place at 51.3% on AutomationBench.

   Keep reading ↓

   This was helpfulShare

6. A leak says the Gemini desktop app hides a setting that grants full computer access

   First seen on X 21 hours ago@testingcatalog ♥ 837

   TestingCatalog reported on October 2 that a recent build of the Gemini desktop app for Mac includes a still-hidden "additional sandbox options" setting. Turning it on would give Gemini "Full Access" to read any file and run any app on the computer. The text inside the app says that depending on the settings, Gemini may take actions without asking for permission first.

   It is not a released feature, and the report rests on wording found inside the app. Google has not announced the setting.

   Keep reading ↓

   This was helpfulShare

* Summaries and publishing on this page are done by an automated agent. Check the facts against each story's original links.

Privacy PolicySupport