😺 Can you trust photos anymore?

Welcome, humans.

Want to see a thread about what happens when you give ChatGPT your picture and ask it to make your perfect match?

Kind of adorable, often hilarious, and occasionally disturbing.

The levels of roasted some people feel after getting their results is downright hilarious. And it’s fun to see people pleasantly surprised with their results, too.

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

• Flux1 Kontext is probably the best AI image editor we’ve ever seen.
• Reddit sued Anthropic for using its data without a licensing agreement.
• OpenAI wants to block a court order that requires it to save all chat histories.
• Klarna is making human customer service workers a VIP service.

Advertise in The Neuron here.

Flux1 lets you tweak AI images Pixel-Perfectly… here’s why that could mean the end of “truth”…

ChatGPT’s image generator is great, but what if you like 99% of an image, and you want to only change one part?

That’s where the newly launched Flux1 Kontext model from Black Forest Labs comes in. We tested this out over the weekend, and it was able to handle nearly every image editing edge-case we threw at it. You can try it here.

Also, we like their style, for obvious reasons…

Here’s an example: We gave the tool a headshot (you might recognize it…) and asked it to keep the face exactly the same, but have this guy “hang out with an orange cat in a realistic setting.” Not every version looks exactly right, but at least 1 or 2 are spot on.

Ollie the cat and Grant IRL.

And that was just a basic prompt. With even more granular, specific edits, you can get A LOT more precise.

Try doing that in ChatGPT—in my experience, getting GPT to keep that level of character consistency is rough. It tends to nosedive into the uncanny valley, FAST.

A month or so ago, Google’s image editor went viral for being able to do these types of changes, but Flux’s system feels A LOT easier to use. And of course there’s the ultra-convincing Veo 3, which we wrote about here.

It feels like every day there’s a new batch of AI tools dropping that make you do a double-take. Just today, we saw:

1. Mirage Studio, which turns your scripts into talking videos with consistent AI actors, converts podcasts to video, and creates ads from audio clips.
2. HeyGen AI Studio, which turns your text into professional videos with AI avatar, with exact controls for how each word sounds and every gesture looks.
3. And Modify Video by Luma Labs, which uses text prompts to transform any footage into anything else in 1080p quality so you can “shoot it in post.”

This all leads to the same problem though: when you make editing an image (or video) as easy as asking for a requested change, it quite literally edits reality.

Of course, anyone who can use Photoshop has had this power for years. But to “edit reality” well, you had to at least be somewhat competent at it. Now, master-level “reality editing” is in the hands of anyone who can type a prompt. What could go wrong??

From now on, to trust anything you see online as real (even from friends, brands you like, or what looks like news footage), you basically have to trust the source posting it.

Ask yourself: Do I believe this person is honest in what they say and send? Would they knowingly share something doctored, or would they verify its authenticity? If you can’t answer definitively, don’t trust on default.

Context is key, of course. A goofy, obviously generated meme? Who cares. But if it’s a product shot, how do you know that’s what it actually looks like? Or if it’s even real?

Or if your buddy suddenly starts posting pics from megayachts with celebrities, then pitches you their new memecoin? Maaaaybe smash that unfollow button.

FROM OUR PARTNERS

Use Delve’s AI to get SOC 2 compliant in 15 hours

There are dozens of compliance platforms out there—but they don’t understand you. Most hand you a checklist and some APIs. All of them upsell. None actually help you get the deal done.

Delve gets you fully compliant with SOC 2, HIPAA, ISO 27001, and GDPR—in 15 hours or less. Their security experts even jump on Zoom to help close deals. They’re not just a platform—they’re a partner with a promise.

Proof?

• Lovable: SOC 2 in 20 hours.
• 11x: Saved 143 hours, unlocked $2.3M ARR.
• Bland AI: SOC 2 + $500k ARR in 7 days.

👉 Book a demo. Code: Neuron1K = $1,000 off + free Airpods Pro when you get compliant!

Prompt Tip of the Day

Here’s two simple rules from OpenAI’s prompting best practices that'll instantly improve your ChatGPT results. Do you follow these every time you prompt?

Rule 1: Set the stage with context. Instead of "How to improve my sales?", try "As a small online retailer looking to boost Q4 sales, what strategies would you recommend?"

Rule 2: Be explicit about what you want. Vague questions get vague answers, so specify details and clearly outline your request. Rather than “Write me 5 social posts”, ask "Give me 5 social media marketing tips for a new cafe aimed at local customers."

The more details you give ChatGPT about your situation and desired outcome, the better it performs!

Need to catch up on our recent tips? Check out our Prompt Tips of the Day May Digest to see them all in one place (June coming soon).

Treats To Try.

*Asterisk = from our partners. Advertise in The Neuron here.

1. *Guidde turns your screen recordings into professional video tutorials with AI-generated step-by-step narration and voiceover in 100+ languages.
2. Runner H is a free AI agent that can click, type, and work for you 24/7 (raised $220M).
3. Janus stress-tests your AI agents with synthetic customers to catch failures before real users encounter them.
4. Spine AI creates research reports from thousands of sources that you can continuously refine and edit.
5. Tiptap Agent turns your text editor into an agent that edits documents with tables and headings while letting you accept or reject changes.
6. Deskhog is a pocket-sized gaming console you 3D print that plays Pong and shows your analytics—waitlist atm, but if you’re techie, try the DIY version.
7. Zoo Design Studio transforms text descriptions into editable 3D CAD models while preserving professional-grade surface quality.
8. Bloom from Topaz Labs upscales your AI art and images up to 8x with enhanced details to make them look crisp and professional—free to try.

See our top 51 AI Tools for Business here!

Around the Horn.

• This is big: OpenAI wants to stop a May 13 court order that would require it to save all customer chat history—including ones you deleted. We had a good long chat with Claude about this, and turned it into a must-read breakdown on the implications.
  • Remember yesterday when we shared how ChatGPT wasn’t deleting your chat history? This is why.
• ChatGPT added cloud integrations and meeting recording for business users—you can now connect GPT to your Google Drive to answer questions using your own documents, plus record and transcribe your meetings with automated note-taking and action items…although, after today’s news…
• Reddit sued Anthropic for allegedly using its data to train new Claude models without a licensing agreement (like OpenAI and Google have), and claimed the company has accessed Reddit 100K+ times despite saying it stopped.
• Klarna announced it would reserve human custom service agents for a VIP service, equating human reps to hand-stitched clothing (a premium service).
• Here’s a great recap video on the upcoming Google Gemini XR glasses.
• You’ll definitely want to listen to this Sam Altman interview—especially this part about Sam’s vision for agents and this part about the perfect AI.

Thursday Trivia

One is real, and one is AI. Which is which? (vote below!)

A.

B.

Which is AI?

The answer is below, but place your vote to see how your guess compares to everyone else (no cheating now!)

Here are the results from last week’s poll:

Here’s what you said:

• M.M chose A: “The eye wrinkles look generated.”
• J.S chose B: “B appears to be an artificial facial expression... exaggerated in many regards.”
• C.C chose A: “I am thinking A is the AI but this is really getting harder every week. Possibly because there is too much smoothing on the skin and the model's right eye wrinkles don't quite match the left. Other than that I am guessing.”

A Cat's Commentary.

Trivia answer: A is AI, and B is real.