Welcome to the BrevFeed daily tech briefing for Thursday, October 1. We've got 14 stories for you today across A.I., security, software, cloud, hardware and startups. Let's get into it.

In AI, Google has unveiled Gemini 4 Argon, its new flagship AI model, which they describe as the next era of frontier intelligence. This model boasts advanced capabilities in coding, reasoning, and especially cybersecurity defense, featuring a significant 1 million output token limit.

Google states that Gemini 4 Argon outperforms top models from OpenAI and Anthropic in several benchmarks, including software engineering and economic analysis. It also shows improved vulnerability discovery over previous Gemini versions, even finding a critical vulnerability in healthcare software.

Access to Gemini 4 Argon is rolling out in phases, starting with trusted cyber defenders and internal Google teams. The company is also participating in the U.S. government's voluntary pre-release model access process before expanding availability to paid API customers, developers, and enterprises.

Pricing for Gemini 4 Argon is set at two dollars per million input tokens and ten dollars per million output tokens. Internally, thousands of Googlers are already using the model for tasks like codebase migrations and optimizing data center memory.

In AI, Anthropic CEO Dario Amodei has proposed a three-step plan to slow down the rapid pace of AI development. His initiative, titled "We Must Pace the Frontier," aims to allow time for safeguards and regulation to catch up with the technology.

Amodei's plan includes granting third-party evaluators access to AI models, establishing common industry safety standards with government involvement, and coordinating global regulatory efforts between democratic and authoritarian governments. Anthropic has already committed to the first step, partnering with Accenture to embed evaluators within their operations.

The proposal has garnered mixed reactions. While figures like Sam Altman and Elon Musk have supported the call for a slowdown, others like Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg are not convinced. China's Foreign Ministry has also criticized the idea, calling it "fear mongering" and a strategy to hinder Chinese competition.

In AI, California Governor Gavin Newsom has issued an executive order to establish state-level AI oversight, citing a lack of federal action. The order directs a group of experts to provide recommendations within two months for strengthening AI safety laws, potentially including requirements for independent safety plans and "kill switches" for frontier AI companies.

Newsom's order suggests measures such as requiring AI companies to embed independent verification groups onsite for regular audits, making transparency reports and risk assessments subject to independent auditor standards, and mandating companies report "loss-of-control incidents" as critical safety events. This initiative positions California to lead in AI regulation.

Additionally, Newsom signed seven bills regarding AI data centers, requiring them to pay for local power grid and water system upgrades, preventing utility costs from shifting to residents. Proposed data centers must also disclose estimated water use, energy efficiency, and drought planning information to local governments.

Finally, Newsom signed SB 947, the "No Robo Bosses Act," which prohibits employers from using AI as the sole or primary basis for firing or disciplining workers. Employers must have a human reviewer corroborate AI-driven decisions and provide written notice to affected employees, including a human point of contact to explain the decision.

In AI, OpenAI has unveiled "Dots," a new type of persistent AI agent, at its recent DevDay event. These agents are built on the GPT-6 Astra model and are designed to operate continuously in the background, pursuing user-defined goals with minimal oversight.

Unlike many existing agents that require constant prompting, Dots are described as "always-on" systems capable of managing multiple projects and retaining information across different platforms like ChatGPT, Slack, and Microsoft Teams. They function using their own cloud computers and browsers, enabling them to perform tasks independently.

Dots can connect to over 4,000 applications through OpenAI's plugin ecosystem, providing a broad range of functionalities. They are available to ChatGPT Pro and Business Premium users, with OpenAI envisioning a future where users can control several Dots, and potentially teams of Dots working collaboratively on their behalf.

Turning to AI and search, new bot controls have been announced to help web creators manage AI's impact on search traffic. This comes as over 50% of online traffic is now non-human, and AI-generated summaries have drastically reduced traditional link clicks, with only 8% of users clicking through when AI summaries are present.

These new bot options aim to give site owners more control over AI access to their content. The measures seek to ensure transparency and uphold existing revenue models that have been disrupted by AI summaries, which often place chatbot responses above traditional links, pushing them further down the page.

Google has also been integrating AI Overviews into its search results, with its VP of search calling AI the most significant upgrade to Google Search. While Google is piloting a program to pay websites for content used in AI results, some publishers report earning very little, while others have seen over a million dollars annually.

In hardware, Apple is introducing a new AI feature for its upcoming Watch Series 12 and Watch Ultra 4. This feature will generate daily recaps of user activities and conversations by utilizing the microphone, GPS, and other sensors.

Apple states that audio recordings will not be stored, and the recaps, which appear in the Siri app, will not include conversation transcripts. The company emphasizes privacy, noting that the feature is opt-in and processes raw audio within a dedicated hardware component called the Secure Exclave, making it inaccessible to the operating system or Apple.

This new functionality requires the latest generation of Apple Watches due to new chips needed for secure audio processing. The Series 12 also includes an improved blood pressure sensor, a redesigned sensor platform for vitals, and a ceramic option. Siri Recap and Live Rewind will arrive in beta with English support later this year as part of watchOS 27.

In a historic move for education, billionaire Ken Griffin has donated three billion dollars to Carnegie Mellon University. This is the largest single donation ever made to a US educational institution.

The massive gift will fund a new 35-acre campus in Miami's Wynwood neighborhood, aiming to enroll over 3,500 students by 2028. This Miami campus will focus on societal challenges like human health, national security, and climate resilience, rather than traditional majors.

One billion dollars of the donation is earmarked for Carnegie Mellon's computer science school, which will be renamed the Kenneth C. Griffin School of Computer Science. The funds will also support the university's Pittsburgh operations, including recruiting academics and increasing financial aid, as CMU looks to rethink education and research in the age of AI.

In AI, the way models are priced based on tokens is becoming a hot topic, with many finding it misleading. This is because different AI models use unique tokenization methods, meaning the same text can result in varying token counts and, ultimately, different costs.

For example, DeepSeek recently cut the price of its V4-Pro model by 75 percent. While this sounds like a big saving, a lower token rate doesn't always mean lower costs, especially with complex AI workflows and agent-based systems that can consume many more tokens than simple chatbots.

However, new solutions are emerging. Research from Writer shows that optimizing the 'AI harness' around foundation models can cut token usage by nearly 40 percent without sacrificing accuracy. This allows companies to deploy AI more efficiently and addresses issues where developers might have over-relied on consuming excessive tokens.

In AI, YouTube is rolling out several new AI-powered tools for creators within its Studio app. These updates aim to help creators analyze content, optimize video performance, and grow their audience.

New features include an expanded Ask Studio, an AI-powered Q&A tool now available on iOS and Android. Creators will also get AI feedback on unpublished video drafts, with suggestions on pacing, structure, and storytelling. Additionally, a new research feed will show trending content, and an AI agent will monitor back catalog videos for optimization opportunities.

YouTube is also introducing dynamic thumbnail generation, allowing creators to generate and test multiple thumbnails for different audience segments. This will help creators find the most effective visuals to attract viewers and improve engagement on their videos.

The Pentagon has launched a new initiative called "Project Meridian" to identify future military capabilities and technologies. Defense Secretary Pete Hegseth announced the project, which aims to look years and decades ahead at how warfare could evolve and ensure the U.S. maintains a technological advantage.

Leading this effort are Elon Musk, CEO of Tesla and SpaceX; Palmer Luckey, co-founder of Anduril; and former House Speaker Newt Gingrich. The project will operate under the Defense Department's CTO, Emil Michael, and will involve a broader team from various sectors.

Project Meridian's focus is on discovering, developing, and fielding advanced weapons and systems for future U.S. troops across all battlefields. The team has 120 days to submit their findings and propose solutions. This initiative comes as the Pentagon also announced plans for a new four-star Autonomous Warfare Command.

In social media news, Reddit is implementing new restrictions on its 'Old Reddit' interface and discontinuing support for RSS feeds, effective November 13th. These changes are part of the company's efforts to combat large-scale data scraping and automated abuse, particularly from AI bots.

Access to 'Old Reddit' will now be limited to logged-in users who have used it in the last six months. New accounts and inactive users will no longer be able to access it, though logged-in moderators are exempt from this recent usage requirement. Reddit previously required all Old Reddit users to be logged in.

The company stated that RSS has become a common surface for large-scale scraping and automated abuse, and 'Old Reddit' has been identified as a significant source of AI scraping. These moves aim to reduce automated traffic and protect user data.

Additionally, Reddit plans to end public API access by March 2027, which will impact third-party tools and moderator workflows that rely on API access.

In government tech, the Trump administration has launched America.gov, a new website designed to streamline citizen interaction with federal services. U.S. Chief Design Officer Joe Gebbia announced that the platform integrates artificial intelligence from Google's Gemini and Elon Musk's Grok, which is developed by SpaceXAI.

The site aims to improve the user experience for the estimated 40 million people who engage with government websites daily. President Trump, who attended the launch event, described America.gov as a "reinvention of your government for the 21st century and beyond."

While the AI chatbot is designed to be user-friendly, not retaining search history or requiring an account, concerns have been raised about potential AI hallucinations and the serious consequences inaccurate information could have for users seeking critical government services.

In AI news, Florida Attorney General James Uthmeier has requested a court order against OpenAI, aiming to prevent ChatGPT from displaying human-like attributes. Uthmeier argues that giving the AI human characteristics deceives users, increases engagement, and aids in data collection.

The motion also seeks to block OpenAI from developing new AI models without third-party approved safety guardrails and to restrict minors' access to ChatGPT. This action is part of an existing lawsuit against OpenAI, which began in June.

The lawsuit and the new motion follow a criminal investigation into a 2025 mass shooting where the suspect allegedly used ChatGPT. Uthmeier stated that the goal is to stop new model development without independent safety guardrails, prevent the harvesting of children's data, and end the practice of portraying ChatGPT as human or entirely safe.

Turning to security, Apple has released urgent updates for iOS, iPadOS, and macOS to patch a critical vulnerability in its CoreGraphics component. This flaw, tracked as CVE-2026-86950, is an out-of-bounds write issue that could allow arbitrary code execution through maliciously crafted files.

Apple stated that this vulnerability may have already been exploited in targeted attacks against specific individuals, particularly those running older iOS versions. The company has since addressed the issue by implementing improved bounds checking.

The updates are available for a wide range of devices, including iPhone 11 and later, various iPad Pro, Air, and mini models, and macOS Tahoe and Sequoia. This incident follows a similar pattern of sophisticated, targeted attacks seen in previous Apple security updates.

That's the BrevFeed daily briefing. We'll be back tomorrow with the stories that matter in tech. Thanks for listening.