Happy Hump Day {{first_name | Toaster}} 🐪 ,

We've gotten good at using AI to catch people slipping. AI can flag suspicious activity, screen resumes, identify plagiarism, moderate content, spot anomalies in financial transactions, and generally notice things a human might miss.

The era of “no one will know I used AI” may be ending as the same AI that is helping us is starting to snitch on us.

Anthropic just announced that new Claude models will embed an invisible watermark into AI-generated text. You won't see it and neither will the reader. But the text can carry a fingerprint that could identify it as AI-generated, even after copying, pasting, and some light editing. The move is partly driven by upcoming EU transparency requirements for AI-generated content.

Claude isn't the only one. OpenAI is also expanding its work around content provenance, including watermarking and Content Credentials for AI-generated media.

LinkedIn has also started cracking down on AI slop — generic, low-effort AI-generated content that looks polished but doesn't really say anything. The platform says its systems are trained to identify generic or repetitive content and reduce how widely it's distributed. In initial testing, LinkedIn says it correctly identified generic content 94% of the time. LinkedIn isn't saying “don't use AI”, but they are encouraging us to have original thoughts again.

Does that actually make the internet better? Yes, we want students to learn instead of outsourcing the work to AI. We want companies to be transparent when AI meaningfully shapes what they're putting into the world. And we probably do want fewer generic, low-effort, AI-generated posts flooding our feeds. The interesting caveat is that LinkedIn hasn't published a precise threshold for how much AI editing is too much.

Maybe transparency builds trust. But maybe, taken too far, it erodes it too. If we start treating “AI-assisted” and “AI-generated” as the same thing, we risk creating a new kind of trust problem. We may end up in a world where everything needs to prove it was made by a human before we trust it.

We spend a lot of time debating whether certain use cases of AI are beneficial. Is it okay to use AI to write your cover letter or marketing copy? Should companies disclose when something was AI-generated? By now, most of us are probably guilty of using AI to write a post or email. So what happens when the idea and perspective are ours, but AI has helped make it sharper and cleaner? The challenge will be figuring out where that line is.

Team Toast 🥂

​Tired of tech conferences that are just a bro-flex fests? Good news: The Toast Summit is nothing like that.

​The Toast Summit is made for women in tech, by women in tech. Four years in, we've built a community that's not waiting around for the future of tech to include us, we're shaping it.

​This year's theme, Shaping the Future of Tech, is exactly that: dream bigger, build smarter, and help design what comes next. Because women can’t change the system unless we are in it.

📍Platform Calgary
🗓️ October 1, 2026
⏰ 8:30am - 6:00pm

Regular tickets sales open August 19th, at 9:00am MST.

You do great work every day. But by the time it counts, at the review, the interview, the raise conversation, most of it has faded. You're left reconstructing a year from memory and underselling what you actually did.

Leeg is a Career Operating System that changes that. A few minutes a day, it turns your everyday work into evidence you can point to. Each day you get a Boost: one small action, 2 to 5 minutes. Name a win. Flag a skill. Set a goal. It all builds into a record you can use for reviews, raises, interviews, and your next move. And your data is yours, following you wherever your career takes you.

Exclusive for Toast community: get 25% off with code TOAST25.