Meta Aims to Make Ads 100% AI by 2026

⚡️ Headlines

🤖 AI

FDA launches agencywide AI tool — The FDA introduces "Elsa," a generative AI tool designed to assist scientific reviewers and investigators by streamlining tasks like summarizing adverse events and generating database code. [Axios]

Character.AI unveils video generation, social feeds — Character.AI launches AvatarFX, enabling users to create animated videos featuring their AI chatbots, along with new social sharing features. [TechCrunch]

Create videos with your words for free – Introducing Bing Video Creator — Bing introduces a free tool powered by Sora that transforms text prompts into short videos, allowing users to bring their ideas to life. [Bing Blogs]

Introducing Mirage Studio — Captions.ai debuts Mirage Studio, an AI-powered platform that generates lifelike videos with human-like actors from text prompts, streamlining content creation. [Captions]

IBM Unveils watsonx AI Labs: The Ultimate Accelerator for AI Builders, Startups and Enterprises in New York City — IBM launches watsonx AI Labs in NYC to co-create generative AI solutions with clients and nurture local AI talent. [IBM Newsroom]

China is gaining ground in the global race to develop AI agents — Chinese startups are advancing in AI by developing autonomous agents capable of handling complex tasks with minimal human input, surpassing traditional chatbots. [Rest of World]

Adobe launches beta version of its Photoshop app on Android — Adobe releases a beta version of Photoshop for Android devices, aiming to attract mobile users and expand its creative tools' accessibility. [TechCrunch]

Codex agent internet access — Simon Willison discusses enabling internet access for Codex agents, exploring the implications and potential applications of this capability. [Simon Willison’s Weblog]

Memory FAQ — OpenAI provides a FAQ detailing how memory works in its models, including how data is stored and used to improve user interactions. [OpenAI Help Center]

🦾 Emerging Tech

Trump crypto wallet — Citation Needed reports on a new cryptocurrency wallet associated with former President Trump, highlighting its features and political implications. [Citation Needed]

Pump.fun token sale — Blockworks covers the launch of Pump.fun's token sale, discussing its unique approach to decentralized finance and community engagement. [Blockworks]

🤳 Social Media

Elon Musk says XChat is rolling out to all, but questions remain about its alleged security — Elon Musk announces the rollout of XChat to all users, though concerns about the platform's security measures persist. [TechCrunch]

Exclusive: TikTok says it has banned 'SkinnyTok' hashtag worldwide amid French pressure — TikTok bans the 'SkinnyTok' hashtag globally following pressure from French authorities over concerns about promoting unhealthy body images. [Euractiv]

🔬 Research

DGM — Sakana AI introduces DGM, a novel approach to generative modeling that enhances the capabilities of AI in creating complex data representations. [Sakana AI]

Clairity Becomes the First FDA Authorized AI Platform for Breast Cancer Prediction – Historic Milestone for Women's Health — Clairity receives FDA authorization as the first AI platform for breast cancer prediction, marking a significant advancement in women's health diagnostics. [Business Wire]

BioReason — The Bo Wang Lab presents BioReason, a tool designed to enhance biological reasoning in AI models, improving their interpretability and application in biomedical research. [Bo Wang Lab]

⚖ Legal

Apple Challenges EU Order to Increase Compatibility With Rivals' Products — Apple contests a European Union mandate requiring greater compatibility with competitors' products, arguing it could compromise user security and privacy. [The Wall Street Journal]

🎱 Random

Airbnb CEO Brian Chesky on services, redesign, app, and the future of travel — Airbnb CEO Brian Chesky discusses the company's app redesign and his vision for the future of travel in a detailed interview. [The Verge]

Employee surveillance software VC funding — Rest of World reports on the surge in venture capital funding for employee surveillance software, raising concerns about privacy and workplace ethics. [Rest of World]

TSMC expects record 2025 profit despite US tariffs, currency volatility — TSMC forecasts record profits for 2025, demonstrating resilience amid challenges like US tariffs and currency fluctuations. [Nikkei Asia]

🔌 Plug-Into-This

Meta Aims to Fully Automate Ad Creation Using AI

Meta is rolling out a new AI-powered ad system that generates entire marketing campaigns—from text to visuals—without human input, targeting small businesses eager for turnkey solutions. This initiative is part of Meta’s broader strategy to embed generative AI into its core business and revitalize its digital advertising engine.

• The system allows advertisers to input basic details (e.g. goals, brand info) and receive fully-formed campaigns tailored to specific audiences.

• Meta’s generative AI can compose copy, generate product images, and create varied ad formats across platforms like Facebook and Instagram.

• The platform supports ongoing optimization, dynamically adjusting ads based on engagement and performance data.

• This marks a transition from assistive to autonomous AI in advertising, with Meta positioning itself as a one-stop AI ad agency.

• The initiative is framed as particularly beneficial for small-to-medium businesses lacking in-house marketing teams or resources.

🤔 Meta wants to be your AI marketing department—just give it your brand and it’ll build your entire ad campaign for you. By collapsing creative, media planning, and execution into a single automated system, Meta is pushing adtech toward a future where data-informed design becomes fully generative—a shift that may further marginalize traditional creative roles.

The Wiretap: Trump Says Goodbye To The AI Safety Institute

The Trump administration will dissolve the U.S. AI Safety Institute (AISI) and replace it with the Center for AI Standards and Safety (CASS), signaling a pivot from exploratory research to formal standardization in federal AI oversight. This move is framed as a step toward regulatory streamlining and industrial alignment.

• The AISI, housed within NIST, was formed to assess AI risks through technical evaluations and benchmarks.

• Its replacement, CASS, is charged with issuing safety and standards guidelines, with a focus on near-term commercial deployment.

• Critics warn this may dilute rigorous risk assessments in favor of industry-friendly frameworks.

• The shift follows broader Republican skepticism toward overregulation of AI, emphasizing “innovation-first” approaches.

• CASS is expected to interface more directly with industry consortia and standards-setting bodies.

🏢 The U.S. is swapping an AI watchdog for a rulebook writer—more cheerleader, less critic. This restructuring echoes a growing trend of softening state-led AI safety interventions in favor of public-private standardization—raising questions about independence, enforcement, and global coordination.

My AI Skeptic Friends Are All Nuts

Thomas Ptacek, co-founder of Latacora, makes a full-throated defense of LLMs in software development, arguing that AI skeptics are missing the quiet revolution unfolding among working engineers. The piece documents how real-world devs are using LLMs to dramatically streamline daily tasks—and aren’t looking back.

• Ptacek describes AI agents that can navigate file trees, write tests, and manage git—all from a chat interface.

• He critiques online discourse for overemphasizing hallucinations, noting developers routinely work around or debug LLM mistakes.

• The post contrasts the rapid assimilation of AI in dev workflows with lingering doubt among pundits and theorists.

• Ptacek draws parallels to earlier paradigm shifts like Stack Overflow, framing LLMs as the next step in programming augmentation.

• While skeptical of AI’s creative applications, he asserts its utility in boring, essential engineering work is undeniable.

🧑‍💻 Developers aren’t debating AI—they’re using it to skip grunt work and write code faster. Ptacek’s grounded perspective captures a turning point: LLMs are becoming invisible infrastructure for engineering teams—used reflexively, not reverently. He received a LOT of hate for it too…

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse