Welcome, humans.

In a recent interview with The Hollywood Reporter, Black Mirror creator Charlie Brooker explained why he avoids too much news:

“One of the reasons why I have to ration my news intake is because I jump five lily pads over to dystopian horror.”

Well, let’s jump with him. Because the latest headlines feel like they’re straight out of his writers’ room. But in this case, reality has already caught up with the script.

An Arizona family recently used AI to generate a deepfake version of a road rage victim delivering his own impact statement in court… after his death.

Meanwhile, NBC Sports is planning to use the AI-generated voice of legendary announcer Jim Fagan in upcoming NBA coverage. Fagan’s been dead since 2017.

And when AI’s not raising the dead, it’s coming for the living. Fiverr CEO Micha Kaufman recently sent a company-wide email warning that AI is coming for their jobs—including his.

Kaufman’s advice? Master the latest AI tools in your field or become obsolete.

So yes, AI is now reviving the deceased, ghosting your career, and—just for fun—breathing life into fictional characters that never existed IRL. Example: Wanna see Pokémon in the flesh? AI’s got you:

Here’s what you need to know about AI today:

• Google Gemini 2.5 Pro I/O Trounces Claude in AI Coding.

• Uber and Robotaxi Collab on Middle East Launch.

• OpenAI is getting intro drug discovery.

• Hugging Face Launches Agentic AI Tool Open Computer Agent.

Google’s new coding AI is now the best coding AI there is… but will the devs finally switch?

Google just dropped Gemini 2.5 Pro Preview I/O Edition (yes, that’s the name; they just keep getting worse and worse), and it’s a turbocharged upgrade to its flagship model. It’s named like it just came off an assembly line at NASA—but performs like a dev’s new best friend.

Here’s the deets: This souped-up Gemini 2.5 is billed as smarter, faster, and cleaner across key benchmarks, giving Google a solid lead in the AI arms race.

For developers, it’s a code-lover’s dream: more accurate function calls (translation: it makes fewer dumb guesses about what your code is supposed to do), faster editing, and creates apps that actually behave like you told them to.

Is it actually that good? Currently, the model leads the WebDev Arena Leaderboard, a benchmark that measures how well an AI can design polished, functional web apps—and for the first time, it edges out Claude.

It also leads in areas like math, creative writing, and answering long-winded queries (so go ahead—dump your entire brain into it and call it “efficiency”).​​

As far as actual anecdotes are concerned, consensus seems to be:

• When Gemini 2.5 is guided and its output carefully reviewed, it can do quite well.

• However, it's not a replacement for expert human judgment, particularly in complex design and architecture.

• Many coders still face challenges with its verbosity, occasional stubbornness, and the surrounding ecosystem (UI, billing, model naming).

Yuchen Jin, co-founder and CTO of Hyperbolic Labs, called the update his top pick for coding tasks.He jokingly floated the name “Gemini 3,” but Google’s marketing team clearly prefers a good acronym over a good sequel.

The “I/O Edition” conveniently launched just ahead of the company’s annual dev conference, where a full parade of AI tools is expected to take the stage. P.S (or should we say, PSA) to whoever needs to hear it: This idea is bad; just call it Gemini 3, y’all!

Our take: The pressure is on. With OpenAI, xAI, and others preparing their own next-gen launches, Google’s now the AI to beat when it comes to coding. Whether or not coders come around and choose Google remains to be seen. Early reports suggest maybe?

Right now, developers are using a portfolio of tools and will switch based on which model best solves their immediate problem or aligns with their workflow preferences, rather than a wholesale allegiance to a single provider.

Gemini 2.5 Pro has definitely made a strong case for itself in certain coding aspects, attracting some users, but it hasn't yet universally "won over" the dev community from established alternatives like Claude or OpenAI. Momentum can be quite powerful, tho…

FROM OUR PARTNERS

See how Sana Agents transform workflows at scale

Sana’s all-in-one AI platform lets you build and deploy powerful agents that automate complex workflows for every team — with no coding experience needed.

With Sana Agents, you can:

• Conduct hours of deep research in seconds across every app

• Analyze complex datasets 10x faster

• Accelerate your sales cycle by cutting prep time 90%

• 100+ pre-built connectors with Slack, Teams, and CRM embeds

• Enterprise-grade security and zero data leakage

Learn more about Sana AI and drive enterprise-wide innovation today.

Prompt Tip of the Day

Here’s a good summary of the above video and how to apply its key takeaways into your own prompts.

Treats To Try.

1. *Incogni removes your personal information from data brokers and people search sites that collect and sell your sensitive data without consent, handling the entire removal process for you.

2. Mistral launched a new cost-efficient enterprise AI model that claims to match 90% of Anthropic's premium model's performance at a lower price.

3. Figma Make is Figma’s new tool for creating websites, app prototypes, and marketing assets, challenging competitors like Canva, Adobe, and WordPress.

4. Anthropic launched an API for AI-powered web search, enabling devs to build apps that tap into current web data without managing search infrastructure.

5. Intercom’s Fin Agent handles customer support across all channels 24/7 so you can focus on growing your startup—free for one year with 300 monthly resolutions.

6. Zed an open-source code editor, gives you an AI agent that helps you write and refactor code; you can chat with it about what you need, and it will suggest or make direct edits to your files for you to review—free personal plan (50 AI prompts/month).

7. Hugging Face has launched Open Computer Agent, a free cloud-based AI that performs tasks on a virtual Linux machine, though it's currently slow (and gets confused more easily than your uncle trying to install Chrome).

See our top 51 AI Tools for Business here!

*This is sponsored content.

Around the Horn.

• OpenAI held talks with the FDA about using AI to speed up drug evaluation processes, with Elon Musk's DOGE associates also participating in the discussions.

• Uber and Pony.ai have announced a strategic partnership to deploy Pony.ai's autonomous robotaxis on Uber's platform—beginning in the Middle East later this year, with plans to expand globally.

• Lightricks just unleashed LTXV-13B, an open-source video generation model that renders 30x faster than competitors—and no, you won’t need a GPU farm to run it.

• Amazon debuted Vulcan, a warehouse robot with a sense of “touch” that can feel objects it handles and is capable of processing 75% of Amazon’s inventory.

• Apple wants to add AI search engines from OpenAI, Perplexity, and Anthropic to Safari as searches decline—the company’s SVP Eddy Cue testified that AI providers will eventually replace traditional search engines, and Google’s stock dropped by 7%.

• If you work in HR, you might be one of the first jobs to be replaced by AI—IBM just said it used AI to automate the work of 200+ HR employees, and used the savings to hire more sales reps and coders.

Thursday Trivia:

One is AI, and one is real. Which is AI? Vote below!

A.

B.

Here are the results from last week’s poll:

Here’s what you said:

• S.C. chose A: “unnatural lighting made it obvious, spelling on sign made it certain.”

• In fact, most people who chose A caught on to the spelling of driveway being wrong. As J.G wrote: “I’m not the best speller person but I’m pretty sure that’s not how you spell ‘driveway.’”

• L.C. chose B: “The clouds are too perfect.” Turns out, it was just a perfect day!

A Cat's Commentary.