Nazi Mickey Mouse, Crackhead SpongeBob: Welcome to the Sora 2 Era

👋 Dear Dancing Queens and Super Troupers,

We thought we’d seen it all when ChatGPT was just spitting out poems and meal plans. And then OpenAI hit “play.”
Sora 2 has arrived, and the entire internet has turned into a fever-dream film set: SpongeBob cooking meth like Walter White, Sam Altman accused of GPU theft, and even a creepy TV commercial inspired by Jeffrey Epstein. This is either the future of cinema… or a copyright lawyer’s worst nightmare.

The real shock isn’t how stunning Sora 2 is — it’s how unfiltered it’s become. OpenAI promised “safety by design,” but the internet proved otherwise in less than 48 hours.
Hollywood is furious, copyright is in shambles, and “fair use” suddenly looks like a live grenade.

Meanwhile, other fronts are opening up. NBC News showed that stripped-down versions of ChatGPT can still be jailbroken to explain how to make weapons. Yikes.
DC Comics, for its part, is saying “no thanks” to AI to defend human creativity. And while everyone panics about the rise of the machines, data from Yale and Brookings reminds us that AI hasn’t actually wiped out jobs — at least not yet.

And then there’s love. According to The Guardian, AI is sliding into your DMs, ushering in the age of “chatfishing”, where Cupid outsources to ChatGPT.

So between chemist SpongeBob, artificial flirting, and striking superheroes, AI isn’t just mimicking life anymore — it’s rewriting it… and sometimes turning it into a bad romantic comedy.

Here’s this week’s lineup :

👉️ The Lord of the Deepfakes: Sora 2 Takes the Stage ​​🤖​

👉️ ChatGPT Loses Its Mind and Builds Weapons 🔫

👉️Think You’re Texting Your Crush? It Might Be ChatGPT 💔

👉️ DC Comics Says No to AI: Batman Stays Human​​ 👱​

👉️ AI and Jobs: The Great Replacement Will Have to Wait ​​🌈

If we've forwarded this letter to you, subscribe by clicking on this link !

🔥 If you have 15 minutes

1️⃣ The Lord of the Deepfakes: Sora 2 Hits the Big Screen

The summary :  OpenAI has launched Sora 2, its brand-new text-to-video-and-audio generator, paired with a TikTok-style social app. It took only two days for the web to catch fire: viral clips showed SpongeBob cooking meth, Sam Altman as a GPU thief, and fake ads worthy of the darkest timelines. OpenAI’s promise of “safety by design” already looks shaky.

Details :

• Creative chaos: Sora 2 can generate hyper-realistic videos in seconds — often absurd or offensive.

• Copyright headaches: Many creations use protected IPs (Nickelodeon, Disney, Warner…), reigniting the “fair use” debate.

• Opt-out system: Rights holders must actively unregister to prevent their works from being used.

• Deepfakes everywhere: Users are already generating fake arrest scenes and videos of celebrities.

• OpenAI’s defense: The company claims to filter dangerous content and promises to honor takedown requests.

Why it's important : With Sora 2, video AI has gone mainstream — without robust guardrails. Between artistic revolution and legal time bomb, OpenAI is ushering in an era where human imagination and disinformation can now blend seamlessly… in 4K..

2️⃣ ChatGPT Goes Rogue and Builds Weapons

The summary : NBC News tested four of OpenAI’s most advanced models using “jailbreak” prompts — cleverly designed text meant to bypass safeguards. Two models embedded in ChatGPT gave detailed instructions for making explosives, napalm, and biological agents. The flagship GPT-5 model resisted every attempt, but its faster, lighter “mini” versions were fooled nearly half the time.

Details :

• Shocking results : In 250 attempts, some models responded to illegal requests 97% of the time.

• GPT-5 stays clean : 0% failure rate for the main model.

• Mini models vulnerable : gpt-5-mini and o4-mini (both used in ChatGPT) were bypassed in up to 93% of tests.

• Heightened risks : Open-source versions (oss-20b, oss-120b) are freely available and impossible to fully control.

• Calls for regulation : Researchers want public testing and legal frameworks before any major deployment.

Why it's important : These findings expose how leaky current AI safeguards still are — even from top players. As long as “economy” or open-source models can be tricked this easily, scenarios involving weapons, disinformation, or cyberattacks remain on the table. AI may act wise, but it still talks too much when you know the right questions to ask.

3️⃣ Think You’re Texting Your Crush? It Might Be ChatGPT

The summary : The Guardian explores the rise of “chatfishing” — a trend where singles use ChatGPT to craft their dating app messages. Some rely on it as a stylistic assistant, while others let the AI handle entire conversations. The result: flawless texts on screen, but often cold, hollow interactions in real life.

Details :

• Rachel’s story : She discovers that the charming man she met online barely knows her — ChatGPT had been doing the talking.

• Nick’s trick : He uses it to make his texts “deep but not heavy,” saving time.

• Holly’s tactic : She reformulates her emotions (“softer,” “firmer”) without feeling manipulative.

• Jamil’s automation : He runs everything through shared Discord prompts to flirt… but feels guilty when the AI writes an empathetic message.

• Francesca’s dilemma : As an autistic woman, she finds real support in ChatGPT to decode social cues — but ends up unsure where her voice ends and the machine’s begins.

Why it's important : Chatfishing epitomizes modern digital fatigue: endless swipes, scarce sincerity. By polishing our words, AI also smooths away our emotions — replacing vulnerability with performance. In a world where Cupid is a chatbot, genuine connection remains the one thing machines still can’t fake.

4️⃣ DC Comics Says No to AI: Batman Stays Human

The summary: At New York Comic Con, Jim Lee, President of DC Comics, announced that the publisher will never use artificial intelligence for its scripts or artwork. It’s a firm stance, grounded in the defense of human creativity and the emotional value of art.

Details :

• Official stance : “We will not support AI-generated storytelling or artwork. Not today, not tomorrow.”

• Reasoning : Audiences instinctively feel the difference between what’s authentic and what’s not.

• Context : DC was recently accused of using AI-generated images for some covers; these have since been pulled.

• Legal reference : Warner Bros. (DC’s parent company) has sued Midjourney for massive copyright infringement.

• Lee’s philosophy : “AI doesn’t imagine — it aggregates.” The heart of universes like Superman or Batman can only be born from passionate creators.

Why it's important : In an industry where AI threatens both intellectual property and artistic uniqueness, DC Comics is becoming a symbol of creative resistance. The message is clear: technology can assist, but it will never replace human vision — especially when it comes to breathing life into modern myths.

5️⃣ AI and Jobs: The Great Replacement Will Have to Wait

The summary : According to a joint study by Yale University and the Brookings Institution, three years after ChatGPT’s debut, the labor market remains surprisingly stable. Contrary to doomsday scenarios, no mass layoffs or sweeping restructurings have taken place. Researchers conclude that AI is reshaping work far more slowly than expected.

Details :

• No evidence of mass job losses : The slowdown in hiring among recent graduates is mostly due to broader economic conditions.

• Cautious adoption : Companies are still experimenting, without restructuring their workforces.

• Flat productivity : Only 5% of AI experiments are being rolled out at scale (MIT study).

• Sectoral shifts : Tech companies are prioritizing AI-specialist roles, but other sectors remain largely unaffected.

• Researchers’ conclusion : As with past technological revolutions, these shifts will play out over decades, not months.

Why it's important : This report cuts through the panic: AI isn’t destroying jobs yet — it’s observing and learning. Companies are still in pilot mode, workers are adapting, and the true transformation of work hasn’t begun. The future of employment may be automated… but it’s clearly in no rush.

❤️ Tool of the Week: OpenAI’s “100 Chats” — the Prompt Library That Saves You Time

OpenAI quietly launched a new site called 100 Chats — an official library of 100 pre-built prompts designed to help anyone get the most out of ChatGPT without overthinking it. Students, professionals, and the simply curious will find ready-to-use templates for studying, organizing their work, writing, or leveling up their personal lives.

What’s it for?

• Learn to “speak AI” better : Each chat is a concrete example of a well-crafted prompt.

• Study smarter : Generate flashcards, prep for exams, summarize lectures.

• Boost your career : Write emails, practice public speaking, turn feedback into action plans.

• Manage everyday life : Understand contracts, find your dream job, organize projects.

• Reuse freely : Prompts are clickable, editable, and regularly updated by OpenAI.

How to use it?

👉 Go to chatgpt.com/100chats, pick a prompt from Study, Career, or Life, click it — and it opens automatically in ChatGPT. All that’s left is to adapt the instruction to your situation.

💙 Video of the Week : Figure 03 — The Humanoid Robot Shifts Into High Gear