Ad
Skip to content
Read full article about: OpenAI sets two-stage Sora shutdown with app closing April 2026 and API following in September

OpenAI is killing Sora in two stages. The web and app version goes dark on April 26, 2026, with the Sora API following on September 24, 2026. OpenAI is urging users to download their content before the cutoff dates. Videos and images can be exported directly from the Sora library.

The company says it hasn't decided yet whether there will be a final export window after those dates. If one happens, users will get an email heads-up. Once all deadlines pass, user data gets permanently deleted. The shutdown also takes down the sora.chatgpt.com platform, which handled image and video generation. Full details are on OpenAI's help page under "What to know about the Sora discontinuation."

Sora's demise is part of a bigger strategic pivot. OpenAI wants to funnel compute toward coding tools and enterprise customers—a play that mirrors rival Anthropic—and a super app rolling ChatGPT and other tools into one package. Sora will stick around as a research project focused on world models, with the long-term goal of "automating the physical economy."

Ad
Read full article about: Google's new Gemini update makes it easy to import memories from ChatGPT and Claude

Google is borrowing Anthropic's memory import approach, letting Gemini users bring over saved reminders, preferences, and full chat histories from apps like ChatGPT and Claude. The process works by copying a suggested prompt into the previous AI app, generating a summary, and pasting it into Gemini, which saves the information in its own context. Users can also upload chat histories as a ZIP file (up to 5 GB) and continue previous conversations inside Gemini. Google is renaming "Past Chats" to "Memory," with the rollout happening gradually.

Google's new memory import feature in Gemini: users copy a prompt into their previous AI app, then paste the generated summary into Gemini. | Image: Google

Anthropic pioneered this approach after OpenAI drew criticism for a military deal Anthropic had turned down on ethical grounds. With users already looking to switch, Anthropic wanted to give them an extra reason to make the move. Both Google and Anthropic rely on the same basic method for data extraction—a simple prompt that asks the existing AI app to output everything it has stored about the user.

Read full article about: Cohere releases open source model that tops speech recognition benchmarks

Canadian AI company Cohere has released "Transcribe," a new open-source model for automatic speech recognition. The company says it claims the top spot on the Hugging Face Open ASR Leaderboard with an average word error rate of just 5.42 percent, beating out competitors like OpenAI's Whisper Large v3, ElevenLabs Scribe v2, and Qwen3-ASR-1.7B. Cohere says Transcribe also delivers the best throughput among similarly sized models.

The chart compares seven speech recognition models with more than one billion parameters. The x-axis shows accuracy as word error rate (WER), where lower values are better. The y-axis shows throughput (RTFx), measuring how fast a model processes audio relative to real time. Cohere Transcribe leads with an RTFx of 525 and a WER of about 5.4, making it both the fastest and most accurate model. NVIDIA Canary Qwen 2.5B follows with an RTFx of 418. Models like OpenAI's Whisper Large v3 and Voxtral Realtime are significantly slower and less accurate.

Cohere Transcribe compared with seven other speech recognition models. Models closer to the upper left corner perform best, meaning faster throughput and lower word error rates. | Image: CohereThe 2 billion parameter model supports 14 languages, including English, German, French, and Japanese. It's available for download under the Apache 2.0 license on Hugging Face and can also be accessed through Cohere's API and the Model Vault platform. Cohere plans to integrate Transcribe into its AI agent platform North in the future.

Ad
Read full article about: Federal judge blocks Trump's ban on Anthropic AI models, calls security risk label "Orwellian"

Anthropic has secured a preliminary injunction against the Trump administration in a federal court in San Francisco. Judge Rita Lin temporarily blocked President Trump's order banning federal agencies from using Anthropic's AI models, along with the Pentagon's classification of the company as a security risk.

Punishing Anthropic for bringing public scrutiny to the government’s contracting position is classic illegal First Amendment retaliation. [...] Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government.

Rita F. Lin, United States District Judge

The dispute traces back to a failed $200 million contract. The Pentagon wanted unrestricted access to Anthropic's Claude models, but Anthropic insisted on guarantees that the models wouldn't be used for autonomous weapons or mass surveillance. Defense Secretary Pete Hegseth then classified Anthropic as a "supply chain risk" - making it the first U.S. company to receive that designation. A final ruling is still pending.

Anthropic leak reveals new model "Claude Mythos" with "dramatically higher scores on tests" than any previous model

Update: The leaked draft blog posts have surfaced online, revealing Anthropic’s plans for a new model class above its existing Opus line. The documents show two possible name candidates, details about a deliberately slow release strategy, and a strong focus on cybersecurity.

Ad

Meta's own supervisory body warns that Community Notes are no match for AI disinformation

Meta’s Oversight Board has examined the planned global expansion of Community Notes. Its conclusion: the system is too slow, too thinly staffed, and vulnerable to manipulation, especially given the growing flood of AI-generated disinformation. In certain countries, Meta should not introduce the program at all.