Skip to content

Trending

Top 10 stories discussed in the past 48 hours · Updated Oct 8, 20:55
NO.01SurgingNo comparison

AI Stocks Fall After OpenAI Tells Investors Annualized Revenue Is Near $50 Billion

OpenAI told investors its annualized revenue was roughly $50 billion at the end of September, below the $68 billion figure widely reported the previous month, and AI-linked stocks including Nvidia, Oracle, and CoreWeave fell after the update. CNBC confirmed the figure from an investor presentation. A person familiar with the numbers said the earlier $68 billion included partner gross revenue, while SiliconANGLE, citing the Financial Times, says the newer figure is based on net revenue, reflecting how OpenAI and Anthropic count cloud partners' sales differently. Accounts of the size of the gap differ: the Financial Times and TechCrunch put it at about $20 billion, and this overview does not reconcile them. OpenAI is weighing a funding round of around $30 billion without a finalized term sheet, and reportedly now targets an IPO in 2027. CNBC's Jim Cramer argued the sell-off shows the value of diversification; this is his opinion.

Latest follow-upAI stocks fall after OpenAI's annualized revenue report comes in lower than expected

SiliconANGLE · AI, CNBC · Technology and 1 more source·7 participants
62
Heat index
NO.02No comparison

Anthropic releases Claude Haiku 5.5, a cheaper small model across AWS, GitHub Copilot and Claude API

Anthropic has released Claude Haiku 5.5, its cheapest and fastest small model, now generally available on the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and in GitHub Copilot. Anthropic says it costs about 75 percent less than Claude Haiku 4.5 for most tasks, at $0.10 per million input and $0.50 per million output tokens for prompts up to 100K tokens, rising to $0.50 and $2.50 above that. Non-default temperature, top_p or top_k values return a 400 error. Benchmark claims come from different sources. Artificial Analysis scores it 43 on its Intelligence Index, ahead of OpenAI's GPT-6 Luna at 38, but reports it uses about 3x more output tokens at max effort. Arena says it debuted at #30 in Code Arena WebDev with 1587 points, up 257 from Haiku 4.5. GitHub says its early testing showed it matched Claude Sonnet 5 on many coding tasks with fewer tokens and steps. In GitHub Copilot it is billed at provider list pricing under usage-based billing, and Business and Enterprise administrators control access through the model policy.

9 sources · 11 participants
54
NO.03SurgingNewNew

SpaceX agrees to buy 800 MHz spectrum from Grain Management, pending FCC approval

SpaceX has announced an agreement to buy a nationwide 800 MHz spectrum portfolio from Grain Management, covering up to 14 MHz of paired spectrum. The purchase is subject to Federal Communications Commission approval and, according to the announcement, is intended to help Starlink become a major US mobile carrier. CNBC reported that shares of AT&T, Verizon, and T-Mobile fell in extended trading after the news. The report describes the stock declines but does not establish their cause beyond the timing of the announcement.

1 source · 5 participants
46

Keep reading No.04–10

  1. 04

    Three fired OpenAI safety researchers dispute misconduct claims in open letterSurgingNew

    Three OpenAI safety researchers, Jasmine Wang, Tomek Korbak, and Mikita Balesni, were fired after OpenAI said they mishandled sensitive information by sharing it with an outside AI safety organization. In an open letter, they deny mishandling information outside established company procedures and warn that their dismissal will deter employees from raising safety concerns. OpenAI says it violated company policies by accessing and handling sensitive company information. An OpenAI spokesperson separately said an investigation revealed a "pattern of misconduct" that goes beyond sharing information with the outside group, and denied the firings were retaliation for safety concerns. The researchers call on OpenAI to keep its public commitments on third-party safety auditing.

    1 source45New
  2. 05

    Anthropic launches free, opt-in OSS Scanner for open-source vulnerability scans with no human reviewSurging

    Anthropic has launched OSS Scanner, a free, opt-in service in which its strongest models, including Claude Mythos, periodically scan enrolled open-source projects for security vulnerabilities. Core maintainers of eligible projects can enroll by submitting a pull request to the company's GitHub repository. Eligibility mirrors OSS-Fuzz's criteria, requiring a "critical impact on infrastructure and user security," decided case by case. Anthropic says the reports are fully model-generated, with no human review or triage. The company says this allows faster and more frequent scanning, but that some reports may be incorrect or invalid. Each report is said to include a proof-of-concept, an explanation, and a suggested fix. In a pilot, 85 of 97 checked critical and high-severity findings met Anthropic's disclosure bar, according to the company's own report.

    2 sources45No comparison
  3. 06

    Anthropic's 2026 Claude Usage Policy update, effective November 12, 2026

    Anthropic has published a 2026 update to its Claude Usage Policy, effective November 12, 2026. According to Anthropic's newsroom announcement, the update mainly clarifies existing rules for longer, more autonomous Claude work rather than adding broad new restrictions. It consolidates bans on deceptive campaigns into a new section, narrows elections rules to voter deception and disruption, explicitly bans weapons-related software and surveillance tools, and tightens requirements for high-risk uses and for models connected to autonomous physical hardware. A post by Mustafa Suleyman, which quotes a post by Andrew Curran, says abusive behavior toward Claude will become a violation of the policy from November 12. That claim comes from social media posts, not from the Anthropic announcement itself, and the reports reviewed do not confirm it independently.

    2 sources39No comparison
  4. 07

    Trump Calls Users of 'Artificial Intelligence' Term 'The Enemy,' Urges 'Super Intelligence'

    President Donald Trump posted on Truth Social that the White House considers anyone who uses the term "Artificial Intelligence" instead of "Super Intelligence" to be "THE ENEMY," according to Gizmodo. The report says Nvidia's Jensen Huang, Meta's Mark Zuckerberg, Elon Musk, and Jeff Bezos have already adopted the "super intelligence" phrasing, and that the ai.gov website was changed to read "SI." The report also says the si.gov domain appears dormant. Gizmodo's account is the only basis for these details, and the report does not describe any policy, formal rule, or consequence attached to the post beyond the rhetorical declaration itself.

    1 source32No comparison
  5. 08

    Google Cloud launches unified Gemini agent for enterprises; consumer release later

    Google Cloud has introduced a unified Gemini agent that plans and completes work across apps and devices, initially offered to enterprise customers in private preview. The agent runs in the cloud, keeps context across devices, and works inside Gmail, Drive, Docs, Sheets, Calendar, Slack, and Microsoft 365 using job-specific sub-agents, according to The Verge. TechCrunch reports it connects to systems such as Jira and any MCP server, writes audit trails attributed to the agent itself, and that Google will bring it to consumers later. SiliconANGLE reports the agent routes each job to a matched model, including Gemini Flash and a flagship frontier model, with Anthropic Claude models also available and other private and open models to follow. Google says flexible spending options are coming online, including per-project spend caps that let companies charge AI costs to departments; TechCrunch describes these as real-time spend caps and multi-model orchestration.

    4 sources31No comparison
  6. 09

    OpenAI adds Intelligent UI to GPT-6 in ChatGPT's Chat tab, rolling out to all tiers

    OpenAI has added Intelligent UI to GPT-6 in ChatGPT's Chat tab, so answers can include custom visuals, interactive elements, and quick task tools. Paid tiers (Plus, Pro, Business, Enterprise) received it first and Free and Go tiers followed, with Enterprise access depending on workplace admin settings. OpenAI says the Work and Codex models are unchanged and users can ask for fewer visual outputs. A Wired reviewer's test produced an annotated diagram, a calculator with sliders, and a clickable seat explorer. Ethan Mollick, who had early access, said it was a welcome change from walls of text. OpenAI says paid tiers run on GPT-6 Sol and Free and Go on GPT-6 Luna; this model split appears only in OpenAI's ChatGPT announcements, not in its other reports.

    4 sources29No comparison
  7. 10

    OpenAI releases hundreds of machine-generated math results, drawing criticism over verification and disclosure

    OpenAI released hundreds of machine-generated math manuscripts, including a paper claiming a proof of the full Birch-Swinnerton-Dyer leading term formula for elliptic curves over the rationals under specific conditions, and critics say the release falls short of verification and disclosure standards. The results come from an unreleased internal model, the same one that produced its earlier Navier-Stokes result, despite a September 29 recommendation from the Advisory Group on Mathematics and Artificial Intelligence (AGMAI) to stop testing advanced math problems on proprietary models. OpenAI spokesperson Lindsay McCallum Rémy said the company aims to empower scientists and seeks to release the model. TechCrunch reports the release did not fully meet AGMAI guidelines: only 10 of 719 manuscripts included chain-of-thought releases, and just 42% of proofs were formalized in Lean. A Cambridge and King's College paper found discrepancies between a natural-language proof and its Lean code for a Navier-Stokes-derived problem. Exponential View, analyzing the release, suggests mathematics could split into vast machine-verified work and a compressed human 'effective theory'. Mathematicians' early accounts, relayed by Ethan Mollick, describe proofs that are hard to read without AI help, with some possibly verified by Lean certificates.

    4 sources27No comparison
How is heat calculated?About this ranking

Heat counts independent accounts and organizations discussing the same story. Duplicates count once, and activity decays with a 24-hour half-life. It measures discussion, not article quality.

The ranking covers the past 48 hours. Trends compare only sources monitored throughout the period; they reflect our coverage, not the whole internet. No trend appears without comparable history.

Source lists show only publicly readable coverage. Participants also include accounts and organizations that contribute to heat alone. Channels from one organization may be grouped, so participant counts may be lower than source counts. Open a story to read its coverage.

Surging
Discussion is growing fast
New
First reported within 6 hours
Rising
Discussion is still growing