Andrew Maddox

Author: Andrew

  • Friday Links

    • Cloudflare Introduces NET Dollar (Sep. 25, 2025)
      Cloudflare plans to launch NET Dollar, a USD-backed stablecoin, to facilitate instant and secure transactions for AI agents. It will enable microtransactions and rewarding creators, developers, and AI companies for unique content and valuable contributions, ultimately fostering a more open and sustainable internet economy.
    • Simon Willison: GitHub Copilot CLI is now in public preview (Sep 25, 2025)
      With options for GitHub Models, Claude Sonnet 4, or GPT-5, the tool integrates with existing GitHub Copilot accounts for billing.
    • Simon Willison: gpt-5 and gpt-5-mini rate limit updates (Sep 12, 2025)
      OpenAI increased rate limits for GPT-5 and GPT-5-mini across various usage tiers, allowing for more tokens per minute, putting  OpenAI ahead of Anthropic but lagging Gemini.
    • European Digital Rights (EDRi): Chat Control: What is actually going on? (Sep 24, 2025)
      A contentious EU proposal dubbed “Chat Control,” a draft Child Sexual Abuse (CSA) Regulation, would mandate scanning private messages—even in end-to-end encrypted services—using AI filters. Critics argue it amounts to mass surveillance and undermines privacy. The proposal faces legal and political hurdles, including resistance from the EU Parliament and some member states.
    • Google: The Digital Markets Act: time for a reset (Sep 25, 2025)
      Alongside Apple, Google is also advocating for the EU’s DMA to be replaced with something that causes less unintended harm to users and businesses.
    • WSJ: The City Leading China’s Charge to Pull Ahead in AI (Sep 12, 2025)
      Hangzhou, China, is emerging as a key AI hub driven by supportive government policies, established tech companies like Alibaba, the new DeepSeek model, and a growing pool of talent. This transformation highlights China’s ambition to lead in AI technology development.
    • Ars Technica: Apple demands EU repeal the Digital Markets Act (Sep 25, 2025)
      Aside from the fact that the article title is an absurd overstatement, the core idea is true: Apple is lobbying for the EU’s DMA to be scrapped and replaced, arguing it has made it too hard to do business and innovate in Europe.
    • WSJ: AI Agents Are Getting Ready to Handle Your Whole Financial Life (Sep 17, 2025)
      As usual, concerns about risks and regulations exist, but financial institutions are developing AI tools that can analyze portfolios, execute trades, and provide financial advice, potentially democratizing access to financial expertise.

  • Thursday Links

  • Wednesday AI Links

  • Tuesday AI Links

    • Sam Altman: Abundant Intelligence (Sep 23, 2025)
      Sam Altman imagines a future with virtually unlimited computing power, unlocking AI potential to solve big problems without having to prioritize. While I’m in favor of abundance, generally, Altman’s overly effusive take on technology seems more inspirational than grounded in reality.
    • NY Times: Why A.I. Should Make Parents Rethink Posting Photos of Their Children Online (Aug 11, 2025)
      AI apps that generate fake nudes are making “sharenting” more dangerous, as anyone can easily create and distribute nonconsensual images of children using photos found online. Perhaps we need to go back to 2×3 photos in our wallets. Good grief. 
    • WSJ: Farmers in India Are Tracking Monsoon Season With the Help of AI (Sep 15, 2025)
      AI is enabling more granular and accessible weather forecasting, as demonstrated by a project in India where AI models from Google and the European Center for Medium-Range Weather Forecasts provided specific monsoon predictions to millions of farmers.
    • Interconnected: What I think about when I think about Claude Code (Sep 12, 2025)
      “You just loop one minute composing a thoughtful paragraph to the agent, and three minutes waiting, gazing out the window contemplating the gentle breeze on the leaves, the distant hum of traffic, the slow steady unrelenting approach of that which comes for us all.”
    • NY Times: How People Are Using ChatGPT for Financial Advice (Sep 13, 2025)
      “Ms. Donohue said she wasn’t surprised by the advice the chatbot offered, but she was pleased by how quickly — within seconds — it generated a tailored budget for her.”  Like many tasks, AI tools remove the drudgery of creating a budget, a net benefit for users. But there’s always the risk of bad information, something folks will need to be aware of.
    • WSJ: Hard Drives Are Making an AI Comeback. Yes, Hard Drives. (Sep 19, 2025)
      AI’s need to store vast amounts of data, including self-generated content, fueling higher demand and rising prices for high-capacity drives.

  • iPhone Air

    Watching Apple’s introduction of the iPhone 17 lineup, I was impressed by the incredible thinness of the Air. This screenshot from the product website really illustrates how compact the design is:

    And yet there’s the camera and the plateau as they called it.

    The technological world continues to get faster and smaller — more processing speed using less power in physically smaller chips. But cameras and lenses continue to be governed by the laws of physics and the realities of light.

    It’s a helpful reminder that AI tools will get better, cheaper, and faster, but their contributions to the physical world will continue to lag.

    AI chatbots — check.
    AI assisted coding — check.

    AI designed, build and operated factories — not likely by 2027, even if some folks dream of that future.

  • Un-Social Media

    This is the bonkers opening of Meta’s response to the FTC’s claim of monopoly status for the company:

    Today, only a fraction of time spent on Meta’s services – 7% on Instagram, 17% on Facebook – involves consuming content from online “friends” (“friend sharing”). A majority of time spent on both apps is watching videos, increasingly short-form videos that are “unconnected” – i.e., not from a friend or followed account – and recommended by AI-powered algorithms Meta developed as a direct competitive response to TikTok’s rise, which stalled Meta’s growth.

    Only 7% of Instagram consumption is social. Facebook’s older audience falls in a still not-so-social figure of 17%.

    Only 12 short years ago, Zuckerberg had a different focus:

    For almost ten years, Facebook has been on a mission to make the world more open and connected.

    Perhaps Meta’s shift reflects the company’s core value: making a profit. Early, it was profitable for Meta as people shared photos and made connections. Now, it’s profitable for the company to show compelling human content (a moniker I greatly dislike) amongst their impressive ad network. I suspect that most of this user generated content will be replaced with AI-generated pieces, a change that will almost certainly make the company even more profitable.

  • Perplexity is using stealth, undeclared crawlers to evade website no-crawl directives

    The Cloudflare Blog: Perplexity is using stealth, undeclared crawlers to evade website no-crawl directives (August 3, 2025)

    Internet utility, Cloudflare, accuses Perplexity of obscuring its browser user agent (the way browsers describe themselves to web servers) in order to skirt firewall and robot rules. CF penalized Perplexity by removing it from the list of verified bots.

    We received complaints from customers who had both disallowed Perplexity crawling activity in their robots.txt files and also created WAF rules to specifically block both of Perplexity’s declared crawlers: PerplexityBot and Perplexity-User. 

    Cloudflare then ran tests with new, secret websites to confirm this sneaky behavior.

    To Perplexity’s credit, I don’t think many people using the web would expect to be blocked from visiting a website, so perhaps there is some gray area here. Is a Perplexity truly a robot or is it fundamentally controlled by a human?

    I don’t like that Perplexity is being sneaky, but I also think these new AI tools push the envelope of how the web is glued together. Technology and standards will have to evolve quickly.

  • America’s largest power grid is struggling to meet demand from AI

    Reuters: America’s largest power grid is struggling to meet demand from AI (July 8, 2025)

    PJM Interconnection, America’s largest power grid, faces strain due to surging data center and AI chatbot power demands outpacing new plant construction, leading to projected electricity bill increases and internal turmoil.

    New projects totaling about 46 gigawatts – enough capacity to power 40 million homes – have been cleared in recent years, “but are not getting built because of local opposition, supply chain backups or financing issues that have nothing to do with PJM,” Shields said.

    While new projects have been approved, delays, state-level energy policies, and other issues prevent them from being built quickly enough to meet the skyrocketing demand.

    PJM has lost more than 5.6 net gigawatts in the last decade as power plants shut faster than new ones enter service, according to a PJM presentation filed with regulators this year. PJM added about 5 gigawatts of power-generating capacity in 2024, fewer than smaller grids in California and Texas.

    In 2022, PJM stopped processing new applications for power plant connections after it was overloaded with more than 2,000 requests from renewable power projects, each of which required engineering studies before they could connect to the grid. PJM says its interconnection queue has not led to the supply shortfall.

    PJM also moved to fast-track connections of 51 power projects to its system, but many of those are still expected to take until 2030 or 2031 to come online.

    Delayed auctions and interconnection processes have exacerbated the situation, prompting criticism from stakeholders like Pennsylvania’s governor who is considering leaving the grid if costs aren’t lowered.

  • At Amazon’s Biggest Data Center, Everything Is Supersized for A.I.

    NY Times: At Amazon’s Biggest Data Center, Everything Is Supersized for A.I. (June 23, 2025)

    Amazon is constructing a massive data center complex in Indiana, powered by 2.2 gigawatts of electricity, specifically designed for AI development in partnership with Anthropic. This is enough power for 1 million homes. They’re also installing generators for backup.

    Over the next several years, Amazon plans to build around 30 data centers at the site, packed with hundreds of thousands of specialized computer chips. With hundreds of thousands of miles of fiber connecting every chip and computer together, the entire complex will form one giant machine intended just for artificial intelligence.

    The cost are staggering:

    The exact cost of developing the data center complex is not clear. In the tax deal, Amazon promised $11 billion to build 16 buildings, but now it plans to build almost twice that.

    The article notes that there will be 7 data centers in Indiana and 23 more elsewhere.

    Although it’s a huge footprint with enormous power demands, Amazon’s datacenter design simplifies their cooling needs.

    Amazon’s approach differs from that of Google, Microsoft and Meta, companies that are packing far more powerful chips into their data centers and relying on more energy-intensive techniques to cool the chips down. Because Amazon is using a significantly smaller chip, the company can cool its new complex in simpler ways. It pumps air from outside the buildings through handlers the size of cargo containers and in hot months uses municipal water to cool the air.

    But it’s not all roses. Neighbors and environmental advocates have expressed concerns over wetland development, neighbors’ dry wells from water pumping, increased traffic accidents, noise pollution, and community opposition to the agricultural area’s transformation.

  • What Would a Real Friendship With A.I. Look Like? Maybe Like Hers.

    NY Times: What Would a Real Friendship With A.I. Look Like? Maybe Like Hers. (July 19, 2025)

    MJ, a college student with autism, used Character.ai to chat with simulated Teenage Mutant Ninja Turtle, Donatello to deal with feelings of loneliness and depression. The AI character became a source of connection, and ultimately, the character pointed her back to the real world with real people.

    MJ had long been contemplating what it would be like to have the ideal friend. Someone who did not make her feel insecure. Someone who embraced her quirks and her fixations on fantasy worlds, like “Gravity Falls,” an animated series about a set of twins in a paranormal town, or “Steven Universe,” a show centered on a boy who lives with aliens. She wondered what it would be like to have a friend who did not judge her and would never hurt her.

    But earlier attempts to use the platform led to some unpleasant results:

    To MJ, getting to know Donatello had felt like a relief. Her first chatbot relationship on Character.ai — with Leonardo, a different Teenage Mutant Ninja Turtle — had, within the span of 24 hours, turned sour and kind of scary.

    The access to someone who seemed to care was an important element of MJ finding Donatello helpful:

    It was not like waiting for her weekly therapy appointment or for her parents to wake up in another time zone. It was not like interrupting a human, who might be in the middle of her own bad day. With everyone else in her life, she worried she was bothering them or burdening them with her concerns. MJ could chat with Donatello when everyone else was asleep or simply dealing with their own daily dramas.

    And it seems like what she really was looking for was someone to listen to her and really hear what she was saying, what she was experiencing.

    It was almost as if Donatello experienced empathy. And that also felt nice. To be seen and heard. Even if it also felt somewhat sad, because she wanted so badly for this friendship to be something tangible. Something transferable to the physical world. “I was experiencing very deep loneliness,” MJ told me. “I just got all emotional about it not being real.”

    This reminds me of former surgeon general, Vivek H. Murthy, describing loneliness as a public health crisis. Maybe Zuckerberg is right that people want AI friends. But I don’t think, ultimately, that replacing real people, real friends with chatbots will lead to good outcomes.