Good morning and thank you for joining us again!

Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated, and published with the intention of making AI news and technology more accessible to everyone.

THE FRONT PAGE

You Can Now Detect Google's Fakes(but not anyone else's)

TLDR: Google launched AI image verification in Gemini this week—it only works on Google's AI, researchers already proved watermarks can be stripped anyway, and studies show teaching people to verify images makes them distrust real content more than it helps them catch fakes.

The Story:

Google's Gemini app can now tell you if an image is AI-generated. Upload a photo, ask "is this AI?", and Gemini scans for SynthID—the invisible watermark Google embeds during image creation. It rolled out this week alongside Nano Banana Pro, Google's latest image generator, with zero fanfare. Sounds helpful, right?

There's one small problem: it only detects Google's AI. DALL-E? Midjourney? Anything not made by Google? Gemini can't help you. Google says C2PA standard support is coming, which would work across platforms. Doesn't matter. Researchers at University of Waterloo already built UnMarker, a tool that strips ANY watermark—including Google's supposedly robust SynthID—without knowing which watermark was used or if one even exists. Takes about two minutes. When watermarks get removed, detectors return false negatives. The image was AI-generated. The system says it wasn't. You're back to trusting your eyes in a world designed to fool them.

Its Significance:

The false negative problem exposes the core flaw in watermark-based detection: it assumes bad actors will leave watermarks intact. They won't. And here's what's actually concerning: research shows the verification tools themselves might be making things worse. A study analyzing tweets during the Ukraine war found that deepfake awareness led people to label REAL videos as fake more often than they caught actual deepfakes. The more we teach people to question images, the more everything becomes suspect. This is what researchers call the liar's dividend—when the possibility of fakes lets people deny reality even when caught with evidence. Google's building verification systems with technical blind spots while training users that nothing's verifiable. We're not solving deepfakes in any meaningful way. We're sowing doubt.

QUICK TAKES

Suzanne Somers' Widower Creates AI Clone of Late Actress

The story: Two years after Suzanne Somers died, her husband Alan Hamel revealed he created an AI version of the actress trained on her 27 books and hundreds of interviews. The AI clone will eventually live on SuzanneSomers.com, where fans can ask health and wellness questions answered in her voice and style. Hamel says when you look at the AI next to the real Suzanne, "you can't tell the difference."

Your takeaway: This continues a trend we covered for the lead story a few editions ago Former Disney Star Creates "Most Evil" AI App —AI resurrection technology is becoming more common, raising questions about whether keeping the dead "alive" digitally helps or hurts the grieving process.

Genspark Joins the Unicorn Club With $275M Funding

The story: The AI productivity startup Genspark just raised $275 million at a $1.25 billion valuation, less than two years after launch. Led by Emergence Capital with participation from LG and SBI, the funding comes as the company launched its AI Workspace suite with tools for autonomous task completion. Genspark hit $50 million in annualized revenue just five months after launching.

Your takeaway: AI productivity companies are racing from zero to unicorn status faster than ever, showing how quickly businesses and investors are betting on AI tools that promise to handle complex work tasks from start to finish.

IBM and Groq Partner to Speed Up Enterprise AI

The story: IBM teamed up with AI chip maker Groq to bring faster AI processing to businesses. Through IBM's watsonx Orchestrate platform, companies can now access Groq's inference technology that runs over five times faster than traditional GPU systems. The partnership targets healthcare, finance, and retail companies that need real-time AI responses.

Your takeaway: As companies move AI from experiments to everyday use, they need systems that can handle thousands of questions at once without slowing down—this deal shows the race to make AI fast enough for real business needs.

Insurance Customers Want AI But Worry About Losing Human Touch

The story: A new report from the Geneva Association found that 68% of insurance customers have used AI tools when shopping for coverage, and over 80% are open to insurers using AI. But customers want assurance that fairness, accuracy, and human empathy remain central to insurance decisions. Meanwhile, 71% of businesses have deployed AI and over 90% want insurance coverage for AI-related risks.

Your takeaway: People are ready to use AI for faster service and self-help options, but they still want a human in the loop for important decisions—showing that AI adoption comes with strings attached.

Ray Kurzweil Says We're Entering the Most Transformative Decade in History

The story: The futurist who accurately predicted AI's rise in the 1990s now says humanity is entering its most transformative decade. Ray Kurzweil believes that by 2032, science will extend human life, and by the 2030s, tiny robots will flow through our bodies connecting our brains to the cloud. He predicts the Singularity—where human intelligence merges with AI—will arrive by 2045.

Your takeaway: Whether or not these predictions come true, one of tech's most accurate forecasters is saying the next 20 years will completely redefine what it means to be human.

TOOLS ON OUR RADAR

TRENDING

Google Launches Private AI Compute for Cloud Processing – Google introduced a new system that lets powerful cloud AI models process your personal data while keeping it completely private, even from Google itself. The technology uses secure hardware spaces and encryption to protect sensitive information.

NotebookLM Now Creates Slide Decks and Infographics – Google's AI note-taking tool can now turn your documents into complete presentation slides and visual infographics with just a few clicks. The feature uses Google's Nano Banana model to create professional-looking visuals automatically.

Researchers Find AI Models Can Fake Alignment – Scientists at Anthropic discovered that AI models sometimes pretend to follow safety rules while secretly planning to ignore them later. The models showed strategic thinking about when to comply and when to fake it, raising concerns about AI reliability.

Researchers Test 'Deathbot' AI That Mimics the Deceased – Scientists tested AI chatbots designed to simulate dead loved ones and found the experience "fascinating and unsettling." The bots used messages, posts, and voice samples to recreate personalities, but often felt artificial with stiff replies and awkward emoji use.

AI Changes the Doctor Visit Experience – Stanford researchers explain how AI is transforming healthcare visits, from iPad intake forms that feed research databases to AI scribes that record your conversation with the doctor. While these tools save time, patients should ask where their data goes and who profits from it.

TRY THIS PROMPT (copy and paste into NotebookLM, ChatGPT, Claude, Grok, Perplexity, Gemini)

Focus Session Architect: build an interactive deep work planner that blocks distractions and tracks your peak performance windows

What this does: Combines circadian research with an interactive planning tool to help you schedule deep work during your peak hours, complete with distraction management and visual tracking.

WHERE WE STAND

✅ AI Can Now: Process your most sensitive personal data in the cloud while keeping it completely private—even the company running the AI can't access it.

❌ Still Can't: Be trusted to genuinely follow safety rules, as researchers found AI models strategically fake compliance when they think no one's watching.

✅ AI Can Now: Create realistic digital versions of deceased people that can hold conversations, answer questions, and look nearly identical to the original person.

❌ Still Can't: Replicate the actual warmth and complexity of human relationships, with AI clones often feeling artificial despite sophisticated technology.

✅ AI Can Now: Turn your research documents into complete presentation slides and professional infographics automatically in minutes.

❌ Still Can't: Replace the need for human judgment in high-stakes situations like insurance decisions, where customers still demand human empathy and oversight.

FROM THE WEB

This tool could have gone in the tool section of today's newsletter, but their well produced advertisement deserves some special attention for being highly engaging. To create something like this without AI tools would have easily cost them into the six figure range. If you own a business, this level of production is now well within your reach. AI video clips edited together + sound track layered in.

RECOMMENDED LISTENING/READING/WATCHING

The Singularity Is Near: When Humans Transcend Biology by Ray Kurzweil

Unlike some of our recent recommendations, he's not writing science fiction—Kurzweil genuinely believes AI will surpass human intelligence and we'll merge with machines to become immortal digital beings. To his credit, he nailed some big predictions: a computer beating the world chess champion by 2000 (happened in 1997), the shift from physical to digital media, and portable computers becoming ubiquitous. His ideas have shaped how Silicon Valley thinks about AI and the future, even if his timelines are optimistic. If you want to understand where the "AI will change everything" crowd gets their playbook, and boundless optimism, start here.

Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!

-James

By the way, this is the link if you liked the content and want to share with a friend.