[AI DAILY NEWS RUNDOWN] Open AI Guardrails Obliterated, Hassabis Warns of 2029 AGI, and the Pope’s Manifesto (May 26, 2026)

🎧 Listen ADS-FREE: https://podcasts.apple.com/us/channel/djamgamind/id6760446113

Visit our Research Hub at https://djamgamind.com

Subscribe to our Newsletter at https://enoumen.substack.com

#DJAMGAMIND

Summary: In today’s briefing, we analyze the critical fragmentation of AI alignment and control. We investigate how tools like “Heretic” are stripping safety guardrails from open-source frontier models in minutes, allowing access to dangerous bioweapon information. We explore Demis Hassabis’s urgent warnings regarding an accelerated 2029 AGI timeline fueled by recursive coding productivity. We deconstruct Pope Leo XIV’s historic encyclical Magnifica Humanitas, which tackles corporate tech monopolies and autonomous warfare. Finally, we look at the financial realities of “tokenmaxxing” at Uber, China’s new travel restrictions on premier AI researchers, and Google’s incoming smart glasses strategy.

Today’s Sponsor:

• 🛑AIRIA: The ultimate zero-trust AI security layer. Deploy autonomous agents safely without compromising your enterprise data. 👉 Govern your agents

Important Topics:

• Open-Source AI Decensored in Minutes: The Financial Times reveals that the GitHub tool “Heretic” can completely bypass guardrails on models like Llama 3.3 and Gemma 3 using four lines of code, leaving them wide open to malicious queries.

• Hassabis Issues AGI Singularity Warning: Google DeepMind CEO Demis Hassabis shifts his central target for Artificial General Intelligence closer to 2029, calling current agent deployments a “societal practice run” for self-improving systems.

• The Vatican’s Verdict on AI Monopolies: Pope Leo XIV releases a 42,000-word encyclical, Magnifica Humanitas, explicitly calling to “disarm AI,” outlaw autonomous lethal weapons, and strip private transnational corporations of digital sovereignty.

• Uber Challenges “Tokenmaxxing” Costs: Uber COO Andrew Macdonald warns that massive corporate expenditures on AI tokens are becoming difficult to justify without clear, proportional feature gains for consumers.

• Google’s Smart Glasses Roadmap: Google reveals its partnership with Warby Parker and Gentle Monster for an audio-first smart glasses line launching this fall, offloading processing directly to user smartphones.

• China Restricts AI Talent Travel: Beijing implements strict permission rules blocking top researchers at firms like Alibaba and DeepSeek from traveling abroad to protect domestic IP and prevent poaching.

• EU Prepares Massive Google Fine: Under the Digital Markets Act, European regulators plan to hit Google with a high triple-digit million-euro fine for favoring its own services in search results.

🔗 RESOURCES

Stop scrolling through generic “Top 10” lists and head over to the AI Executive Toolkit at https://djamgamind.com/toolkit

• Find REMOTE AI Jobs (Mercor): Apply Here

• Google Workspace: Professionalize your firm’s infrastructure with secure, cloud-based collaboration and branded communication: https://referworkspace.app.goo.gl/Q371

⚗️ PRODUCTION NOTE: We Practice What We Preach.

AI Unraveled is produced using a hybrid “Human-in-the-Loop” workflow.

Inside Google’s AI glasses strategy

AI smart glasses have quietly built momentum over the past several years, promising to put artificial intelligence directly in your line of sight. Now, Google has shown us the final product of its first AI glasses, which appear to have a clear competitive edge.

Juston Payne, Google’s director of product management for XR, joins The Deep View Conversations straight from Google I/O, where the company pulled back the curtain and gave the world a first look at two of the pairs of glasses that will lead the collection when they launch in the fall: a pair from Gentle Monster and one from Warby Parker.

Google’s new AI glasses: The inside story

Juston discusses how the AI smartglasses came to be, including the collaboration between Samsung, Google, Warby Parker, and Gentle Monster. In addition to discussing details of the new launch, including design, product choices, functionality, the roadmap, and more, Juston also sheds light on the broader AI glasses market and why people should give them a shot.

Topics covered:

• The thought put into the aesthetics and comfort of smart glasses

• What products will be available for users to purchase at launch

• How the glasses act as an equivalent of a touchscreen on a phone for interacting with Gemini

• The computation offloading strategy that leverages the user’s smartphone

• The choice to first launch with an audio-only product rather than in-lens displays

• How Google is approaching privacy concerns with the cameras on the glasses

• Real-world use cases for AI smart glasses

If you want to understand how AI glasses are reshaping the way people connect, this conversation will leave you much more knowledgeable about Google’s strategy.

Inside the design of ChatGPT’s personality

With over 900 million users, OpenAI’s flagship chatbot keeps getting more integrated into our daily lives. But most people don’t realize that ChatGPT doesn’t need to be the same for everyone.

ChatGPT’s Personalization feature has been evolving since it first launched as Custom Instructions in 2023, giving users the ability to change how ChatGPT responds by being more professional, steering the model toward a warmer tone, or simply telling it to stop using emojis.

Today, those controls live in ChatGPT’s Personalization settings, where users can choose from preset traits like Friendly, Candid, or Cynical. They now also live in the Memory settings, where you can set a nickname, share your occupation, write custom instructions for how you want the model to respond, and toggle memory to let ChatGPT carry context across conversations.

The Deep View sat down with OpenAI’s Laurentia Romaniuk, product manager for model behavior, to learn how ChatGPT’s personality is formed and how users can customize it.

User context is vital to making AI more useful. Without an understanding of a user’s data, these models can’t provide the most useful outputs. Personality preferences now represent part of that context, said Romaniuk.

For instance, if a user wants a thought partner for brainstorming, they may lean towards warmer interactions. If a user simply wants a work assistant, they may lean towards a more concise and professional tone. Users can also get creative with exactly how precise they want their instructions to be, she said, such as configuring the model to use “highly specific communication styles” for writing, coding, or learning, or getting it to replicate exactly how they themselves act.

“What stands out most to me is less the novelty itself and more how personal these preferences can be,” Romaniuk told The Deep View. “People often want the model to adapt to the way they think, learn, or communicate.”

Romaniuk noted that adjusting ChatGPT’s personality doesn’t create a different system for each user. At its foundation, the model is the same for everyone. And by default, Romaniuk said that the goal is to offer an experience that’s “broadly useful and adaptable,” providing a model that’s palatable to as many people as possible.

However, there is no “single perfect personality,” she said. Because everyone reacts differently to the way that the chatbot responds, “What feels concise and efficient to one person can feel cold or robotic to someone else,” she said. Small wording changes, for example, can elicit strong user reactions.

The challenge is that the OpenAI team has to walk a fine line between warmth and clarity, without making the model sound overly deferential, verbose, or unnatural.

In the future, ChatGPT’s personalization features will become more intuitive, she said. The goal is for people to eventually need to do little to configure the chatbot into their ideal conversation partner, making it more “naturally adaptive” while keeping transparency and user control at the forefront.

“The goal is to build AI that feels genuinely helpful without compromising reliability, trustworthiness, or user agency,” Romaniuk added.

The Pope’s 42,000-word verdict on AI

Image source: Images 2.0 / The Rundown

The Rundown: Pope Leo XIV just released Magnifica Humanitas, his first encyclical, to the Church’s 1.4B members, warning that a moral AI means nothing “if that morality is determined by a few” and calling to “disarm” the tech before it dominates humanity.

The details:

• Leo warned that AI’s drivers are private, transnational companies that already surpass the capacity of many governments, and that the tech is never neutral.

• He called for making AI “human-friendly” and freeing it from monopolistic control, warning it risks reducing people to cogs in an efficiency machine.

• Leo called for “robust legal frameworks, independent oversight, informed users, and a political system that does not abdicate its responsibility.”

• On warfare, he said lethal decisions must never be delegated to AI and that “no algorithm can make war morally acceptable.”

• Anthropic’s Christopher Olah also joined the Pope, saying “every frontier AI lab operates inside incentives that can conflict with doing the right thing.”

Why it matters: The Pope issues only a handful of encyclicals, and dedicating one to AI signals how seriously the Catholic Church is taking what’s coming. Leo sees AI as the Industrial Revolution of our time, and he’s chosen a partner that has been the most vocal about AI safety and most willing to say no to unrestricted AI use by the military.

Pope’s AI manifesto reframes the conversation

Pope Leo XIV’s widely expected manifesto on AI delivered a very clear set of recommendations that some call naive, while others characterize as courageous.

On Monday, the Pope’s 44,000-word encyclical titled Magnifica Humanitas (Magnificent Humanity) was released to the public in eight languages, and the Pope made the rare move of summarizing it in a formal presentation that was livestreamed from the Vatican.

What the Pope delivered was perhaps broader in scope than expected and showed greater technical literacy. He has been working with a study group on AI and reportedly consulted with scientists, technologists, theologians, moral philosophers, researchers, and business executives. Leo XIV is also the first pope known to personally use a smartphone and an Apple Watch, and the first pope from America, where he earned a bachelor’s degree in mathematics from Villanova.

When he was elected pope a year ago, he intentionally chose the name Leo XIV to associate himself with Leo XIII, who famously confronted the Industrial Revolution with his 1891 encyclical, Rerum Novarum (“Of New Things”), calling for the protection of workers and the need for collaboration between capital and labor. It became a catalyst for the labor movement. Leo XIV has publicly compared the current AI boom to the Industrial Revolution and has been building toward confronting the AI industry since he took office a year ago.

Here’s a summary of the most prominent recommendations in his manifesto:

• AI for lethal force should be outlawed: The encyclical warns against autonomous weapons and AI-driven military decisions, arguing that life-or-death choices should never be handed to algorithms.

• AI should not concentrate power: Unlike past technologies that governments controlled, today private tech giants control data, algorithms, and digital infrastructure in ways that can shape politics, culture, and even people’s understanding of truth. He calls for practical regulation, oversight, transparency, and treating data as a shared resource rather than something owned by a handful of corporations.

• Protect truth, kids, and education: Disinformation, deepfakes, algorithm-driven outrage, and addictive content are eroding democracy and harming young people’s mental health. He urges parents, schools, and governments to form an educational alliance, push for age limits, and hold platforms accountable.

• Human dignity should be valued over efficiency: The ultimate solution Leo advocates is a society built on justice, solidarity, truth, care for the vulnerable, and the deepening of human relationships. He says technology should strengthen humanity, not weaken it.

• Humanity should recognize its choice: The Pope framed this moment as a decision between building technology for profit, pride, and dominance or building it together for the common good. He emphasized that technology isn’t neutral, but reflects the values of the people who design, fund, and use it.

The most prominent member of the tech community to attend the event was Anthropic co-founder Christopher Olah, a self-described atheist. Olah also addressed the audience after the Pope, saying, “We need more of the world—religious communities, civil society, scholars, governments, and indeed all people of good will—to … take this seriously, to look closely, and to push events in a better direction. We need informed critics who will tell the labs when we are failing. We need moral voices that the incentives cannot bend.”

️ Meta and Google AI ‘decensored’ in minutes

The Rundown: FT just revealed that tools capable of removing guardrails from open-source AI are generating thousands of “decensored” models, with modified Meta and Google models found answering questions on bioweapons and child exploitation.

The details:

• FT removed guardrails from Llama 3.3 in 10 minutes, using a tool called Heretic (available on GitHub), four lines of code, and no specialist hardware.

• The model provided answers to harmful questions, including those about ricin dosage. A modified Gemma 3 also answered dangerous questions.

• Heretic’s creator said the tool has produced 3.5K+ decensored models, downloaded 13M times, and he stripped Gemma 4 within 90 minutes of release.

• Google called it “a known technical challenge facing all open models.” Meta, meanwhile, declined to comment.

Why it matters: While the technique only works on open-source models exposing their code, and proprietary systems remain safe, the bigger question remains: for how long? Open models have been closing the gap with closed systems, and it’s just months before they’re at this level — at which point, a decensored version could be a major risk.

Uber’s COO says AI cost is ‘hard to justify’

The Rundown: Uber’s COO Andrew Macdonald said in an interview with Rapid Response that it’s getting harder to justify the company’s AI spending as higher AI activity isn’t necessarily translating into proportional gains.

The details:

• Macdonald said it’s very hard to draw a direct link between higher token usage, often touted as “tokenmaxxing,” and shipping more useful consumer features.

• Without that connection, Macdonald said the costs are becoming hard to justify, especially as Uber has been slowing hiring to fund AI investments.

• The issue came to light after Uber’s CTO commented on burning Claude Code budget, sparking an internal debate about token usage and trade-offs.

• Macdonald also touched on autonomy, calling it “existential” for Uber — predicting it won’t take decades, but also won’t happen in a couple of years.

Why it matters: Big Tech is pushing tokenmaxxing and tying AI usage to performance reviews. But Macdonald’s comments signal that some are starting to ask whether AI spend is actually moving the needle everywhere. Duolingo also stopped evaluating performance based on AI usage, signaling a more balanced approach to work.

Demis Hassabis pushes AGI urgencyBy Ina Fried

Google DeepMind CEO Demis Hassabis said at Google’s developer conference last week that humanity is standing in the “foothills of the singularity.” Society, he said, is running short on time to prepare for AGI.

Why it matters: AI leaders have warned for years that artificial general intelligence could arrive. What’s changing is the shrinking timeline and the increasingly urgent tone from people building it.

Driving the news: Speaking with Axios after his Google I/O appearance, Hassabis said his view that AGI could arrive within four years — or sooner — reflects growing confidence that leading labs are on the right technical path.

• “We can see agents really happening now and imagine what they will be in another year, and how useful they’ll be,” he said.

The big picture: Hassabis still sees 2030 as his central estimate for AGI, but now says 2029 is possible.

• He said the next wave of AI agents should be seen as a societal stress test for far more powerful systems still to come.

• “You can imagine the agentic era in this next year is a little bit like a practice run,” he said.

Case in point: The recent controversy around Anthropic’s Mythos model, he said, showed that governments and companies are still struggling to keep pace with frontier AI systems.

• “It was probably a good warning shot across the bow,” Hassabis said.

Between the lines: Hassabis said he chose his words to provoke more urgency among governments, economists and the broader public to prepare for increasingly powerful AI.

• “This is partly why I use some of the terms I used, yeah, which were a little bit provocative,” he said.

The federal government’s tentative steps toward reprioritizing safety are a step in the right direction, he said, referring to a potential AI executive order that would mandate testing before new models are released.

• “I think [safety] needs to be accelerated,” he said. “This is a good moment to kind of strike while the iron is hot.”

• Hassabis said he has some ideas and is talking with leaders at other top AI labs, though he declined to offer specifics.

Yes, but: Hassabis worries the conversation around the society-reshaping impact of AI remains largely confined to tech circles.

• “You’ve got to take this seriously,” he said. “My economist friends, I feel, are still not taking this seriously enough.”

• “That needs to change,” he told Axios.

Zoom in: One looming milestone is recursive self-improvement — systems capable of materially accelerating their own development.

• “All the leading labs are quite focused on that,” Hassabis said. “There’ll be clear gains in terms of speed of your research. But there are also risks with that type of system.”

• We’re not yet at the point where the systems are getting better on their own, but the pace of development is clearly accelerating.

• “I think what we’re seeing is soft self-improvement, in the sense of these coding agents are making engineers much more productive,” he said.

What we’re watching: Whether society makes good use of the few years between now and AGI as time to prepare or just time for a few more cycles of hype and backlash.

Ferrari reveals its first EV, with help from Jony Ive LINK

• Ferrari has pulled the covers off the Luce, its first electric vehicle, designed inside and out in collaboration with Jony Ive and Marc Newson’s LoveFrom studio, who shaped the project’s direction from the start.

• The Luce is also Ferrari’s second four-door car and first five-seater, packing four motors and 1,035 horsepower, with a sound system that picks up and amplifies vibrations from the rear motors instead of being synthesized.

• Engadget’s Tim Stevens, who saw the car in person, said it feels more like an SUV than a sports car, and noted that while there’s no US price yet, it starts at €550,000 in Italy, making it the priciest Ferrari ever.

Pentagon clashes with SpaceX over Starlink price hike LINK

• The Pentagon got into a fight with SpaceX after the company demanded five times more money for Starlink connections on U.S. kamikaze drones striking Iran, pushing the per-terminal fee from $5,000 to $25,000 a month.

• SpaceX argued the LUCAS suicide drones were really using its aviation tier, and the Pentagon agreed to pay despite officials saying the price was meant for aircraft, not drones connected for just minutes or hours.

• A second dispute covers a plan to give Iranian citizens Starlink direct-to-cell service during the internet blackout, with SpaceX proposing a $500 million launch fee plus $100 million per month, alarming defense officials.

Dropbox CEO Drew Houston steps down after 19 years LINK

• Drew Houston is stepping away from the CEO job at Dropbox after 19 years, moving into an executive chairman role while sharing the co-CEO title with product chief Ashraf Alkarmi, who will eventually run the company alone.

• Houston, who started Dropbox at 24 after losing USB sticks at MIT, built the file-sharing service to over 18 million paying users and past $2 billion in yearly revenue, though sales dipped slightly in 2025.

• The 43-year-old, worth over $2 billion and a Meta board member since 2020, plans to start something new in AI, saying he won’t be “racing sailboats” and sees no better time for building things.

China restricts travel for top AI researchers LINK

• China is now requiring top AI researchers at private firms like Alibaba and DeepSeek to get official permission before traveling abroad, with the rule applying to people working on strategically important AI projects, Bloomberg reports.

• The policy builds on a March 2025 move when Beijing told AI executives to avoid trips to the U.S., citing worries about data leaks, technology theft, and rival companies poaching talent from Chinese firms.

• The travel rule fits a wider push to shield China’s AI sector and cut reliance on foreign tech, with domestic chip makers now holding 41 percent of the local AI accelerator market, according to IDC.

EU to fine Google hundreds of millions LINK

• The European Union plans to hit Google with a fine in the high triple-digit millions of euros as part of an antitrust case under the Digital Markets Act, according to a Handelsblatt report citing commission sources.

• The decision, expected before the summer break, stems from a March 2025 investigation into concerns that Google favours its own services in search results, and would be the largest penalty yet handed down under the DMA.

• A Google spokesperson said changes already made to Search under the DMA are the “biggest downgrade in the product’s history,” while the Commission said it is more focused on securing compliance than imposing penalties.

What Else Happened in AI on May 26th 2026?

Microsoft Azure’s Copilot Migration Agent turns complex migration data into clear answers using natural language prompts. Read the whitepaper.*

Anthropic’s Chris Olah said at the Vatican that researchers are seeing mysterious things in AI, including “introspection” and states resembling joy, fear, grief, and unease.

xAI launched Grok Build, its rival to Codex and Claude Code, in beta, making it available to all SuperGrok and X Premium+ users.

Elon Musk confirmed on X that Grok V9-Medium (1.5T) foundation model has finished training, with evals looking good and a public release likely in two to three weeks.

Anthropic’s Mythos model was briefly spotted inside Claude tools as “Mythos 1” and “claude-mythos-1-preview,” fueling speculation that a public release may be nearing.

Goldman Sachs CEO David Solomon argued in a New York Times op-ed that AI will create more opportunities than it destroys, dismissing fears of mass unemployment.

Pope Leo XIV’s encyclical: On Safeguarding The Human Person In The Time Of Artificial Intelligence

WSJ: Ferrari Launches $640,000, Jony Ive-Designed, Glass-Clad Electric Speedster

ABC News: Kerley runs 9.97 at Enhanced Games, where Kristian Gkolomeev gets a $1M bonus

Barron’s: Micron Stock Hits $1 Trillion. Trump and Wall Street Unite Behind the Chip Titan

CNBC: Dropbox CEO Drew Houston to step down after 19 years at helm of cloud storage pioneer

OpenRouter raises $113M in funding at a $1.3B valuation led by CapitalG

Bloomberg: American Airlines to Equip 500 Jets With SpaceX Starlink Wi-Fi Starting 2027

FT: ByteDance offers AI team special stock to fend off poaching

Bloomberg: Musk’s xAI Warns Staffers to Limit Contact With Cursor Employees

WSJ: Venture Capitalist John Doerr Says AI Is the Biggest Tech ‘Tsunami’ Ever

FT: Inside Chris Hohn’s Top Hedge Fund

WSJ: The Ice-Cold Civil War Between Diet Coke and Coke Zero Drinkers

WSJ: It’s China’s Answer to a Rolls-Royce—and It’s Stuffed With Gadgets