All Your Claude-Generated Text Will Soon Include AI Watermarks 👀

Welcome to another edition of Horizon AI,

Anthropic just announced that invisible AI watermarks will soon be added to all Claude-generated content, a change that's stirring up mixed reactions online. In today's issue, we'll look at how the watermarks work, what content they affect, and what this could mean for anyone using Claude to write or edit content.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Anthropic to Watermark Claude's Output Worldwide

• Meta Launches Muse Glimmer Model as Zuckerberg Doubles Down on AI for All

• AI Tutorial: Find new career paths with Google AI

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

ANTHROPIC

Anthropic to Watermark Claude's Output Worldwide

Anthropic announced that new Claude models launched after August 2 will automatically embed an "imperceptible watermark" into any text or file that's passed through it, whether Claude created it or just touched it up.

Details:

• The watermark doesn't change the text's meaning or readability, but travels with it when copied and pasted, and "may persist through some editing."

• Supported file outputs (such as .png, .jpg, .svg) will also include digitally signed provenance metadata using the C2PA standard, which signals that a file was processed by Claude.

• The company said it will extend support for older models as well.

• The watermarking mandate stems from the EU AI Act, which includes a provision requiring that AI systems creating synthetic content mark their outputs.

• But, despite being introduced in response to the EU AI Act, Anthropic said the markings will apply worldwide and across all Claude products, including Claude Cowork, Claude Code, and the Claude API.

Anthropic also plans to release verification tools that will allow anyone to detect the watermarks, with details on those tools slated to arrive in "forthcoming" announcements.

Source

TOGETHER WITH Q3 ADVISORS

The #1 Roth Conversion Mistake That Costs Retirees

Most people missed the first Roth conversion window. History is offering a rare second chance — but not a long one.

With today's tax rates extended, IRA Millionaires have a brief stretch of historically low brackets left. When it closes, the math changes: higher rates, bigger required distributions, and fewer conversion opportunities each year you wait.

Craig Wear, CFP®, has spent 15+ years helping more than 3,000 families with $1M+ in IRAs plan Roth conversions — helping them avoid a projected $10B+ in lifetime taxes. His new book, Roth Conversion Reset, is the step-by-step playbook:

Lock in today's low rates
Avoid RMD surprises at 73
Protect your spouse from survivor taxes
Keep more for your heirs

The book is free — no credit card, no shipping. The full PDF lands in your inbox in minutes.

Get My Free Copy

META

Meta Launches Muse Glimmer Model as Zuckerberg Doubles Down on AI for All

Meta has released Muse Glimmer, a new open-weight AI model designed to run agents directly on consumer computers, offering a glimpse into Mark Zuckerberg's vision of AI that works privately and continuously on behalf of individuals.

Details:

• Muse Glimmer has 30 billion parameters and is available under the Apache 2.0 license, allowing developers to download, modify, and run the model on their own hardware.

• The model is built for multi-step agentic tasks, including using tools, writing and debugging code, working with files and screenshots, and completing longer workflows with limited supervision.

• Glimmer is designed to run locally on a Mac or PC with a single consumer GPU, supporting both text and images, and more than 100 languages.

• Mark Zuckerberg also announced that the company plans to soon open the weights for its latest and more powerful AI model, Muse Spark 1.2.

Glimmer is designed to be "always-on" and able to operate "anywhere, anytime, with or without an internet connection," mirroring the future Zuckerberg has previously laid out for Meta. On Monday, he reiterated that vision in a new letter, arguing that distributing superintelligence widely "has the potential to begin a new era of personal empowerment where individuals can use this powerful new capability to reach their full potential, pursue their interests, and improve their lives and the world more than ever before."

Source

AI Tutorial

Find new career paths with Google AI

Google’s experimental AI tool, Career Dreamer, makes career exploration easier and more personalized using real job market data. By analyzing your background, past roles, and personal interests, it pinpoints your transferable skills, crafts a personalized career identity statement, and helps you explore matching job roles.

1. Go to grow.google/careerdreamer and enter a current or previous job title, along with the industry (optional).

2. Career Dreamer will generate a list of common tasks for your role. Select the tasks you performed, then click Next.

3. It will then draw on real job market data to break down those tasks into transferable skills. Hover over the generated skills to understand why the AI included them, and select those that apply to you.

4. Fill in the building blocks on the left panel (Experiences, Education, Skills, and Interests). You'll be able to unlock additional skills as you fill in more info. You can also add personal hobbies to further personalize your profile.

5. Generate and refine your Career Identity Statement: Review the statement Career Dreamer drafts on the right panel. Click Re-generate to explore different phrasings, or view Version History to restore previous drafts. Once satisfied, copy and paste it into Google Docs for final edits.

6. Click Explore Paths to see a variety of career possibilities tailored to your career identity. Use the filters (in the center) to prioritize roles based solely on skills and interests, experience, and so on.

7. Click on specific roles and select "Learn more" to see salary data and degree requirements. You can navigate down to see more details, like a "day in the life" summary, skill overlap, and areas for growth.

8. Click Find jobs near you to view active local job listings.

9. You can also click Jump to Gemini to access ready-to-use AI prompts for drafting cover letters, updating your resume, or discovering more potential career paths.

AI Tools to check out

🔇 Clumi.ai: Online AI vocal remover.

⭐ Seedeo: Create videos, images, music, voices, and marketing content faster with AI.

✅ Oqoqo: Build evals and custom benchmarks for real-world tasks.

📹 TapVid: An AI explainer video generator that turns a prompt, PDF, or link into a finished video with motion graphics, no editing or design skills needed.

👉 Paritok: Compression gateway for AI coding agents. Spend up to 85% less and run 3× longer coding agent sessions.

AI Findings/Resources

💻 Google Course: Learn how to build custom, time-saving apps from scratch by “vibe coding”

😅 What happens when candidates are all using ChatGPT for their campaigns

🚀 Evolve your marketing with Google's new AI and agentic experiences across Google Ads and Google Analytics

📱 This guy lost his phone at the office, so he asked Claude for help, and it ended up writing him a tool to track it down

😂 A man went viral after trying to meditate with a group of AI chatbots that just wouldn't stop talking

The latest in AI and Tech

OpenAI’s AI ethics chief leaves the company less than a year after joining

Chloé Bakalar, who joined OpenAI as AI ethics lead in August 2025 from Meta, has left the company. Her departure comes alongside several other senior safety and alignment exits.

Anthropic model makes progress on the Riemann hypothesis

An unreleased Anthropic model has made significant progress on the Riemann hypothesis, one of mathematics’ most famous unsolved problems. The model tested 650 ideas using 60 subagents over 31 million output tokens, with its results later confirmed by Anthropic mathematicians and formalized in Lean.

OpenAI expands Daybreak cyber defense service

The company now offers two tiers, Blue and Red, offering tools for incident response, malware analysis, security testing, and vulnerability research. The Red tier also introduces GPT-5.6 Cyber, a model built off of GPT-5.6 Sol, which offers enhanced capabilities for certain specialized cybersecurity tasks.

Spotify to distinguish AI artists from real people

Spotify will begin adding “AI Persona” labels to artist profiles that represent AI-generated identities, with the first badges appearing in September. Music from labeled AI Personas will be excluded from editorial and algorithmic recommendations by default, unless users choose to follow those artists.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻