Hi Reader, we recently hacked MKBHD’s locked iPhone and “stole” $10,000. You can check out how we did it here. No, we didn’t trick him for his password or fingerprint. We used a small device hidden near a payment terminal, intercepting the invisible signals phones send when we tap to pay. This hack has been public knowledge since 2021 and it still isn't fixed. It raises the question of who's in control of the technology we're building, a theme we continue to explore in this week's newsletter.

Coming up this week
🤖 AI race enters new dangerous phase​
​🚀 NASA finds new building blocks of life on Mars
👽 NASA's final plan to save Voyager 1
+ More

ARTIFICIAL INTELLIGENCE

AI Race Enters New Dangerous Phase

Earlier this month, Anthropic built an AI model it considered too dangerous to release. During testing, the model called Mythos autonomously found and exploited security vulnerabilities in every major operating system and web browser, including a 27-year-old hidden flaw in software specifically designed to be secure. Think of it as an AI that can break into almost any computer system, on its own, with minimal human direction. Anthropic shared it with a broad coalition of trusted companies and organisations - including Google, Apple, Microsoft, Amazon Web Services, JPMorgan Chase, and over 40 others - to help patch vulnerabilities before similar tools reached the wrong hands.

Due to its capabilities and potential for misuse, Anthropic decided not to release Mythos. However, there's some speculation that a group of users on a private online chat accessed it through an insider, and have had continuous access since.

Then, last week, Chinese AI lab DeepSeek launched two preview versions of its newest large language model, DeepSeek V4, a much-awaited update to last year's V3.2 model. The company claims its new model outperforms and outstrips OpenAI's GPT-5.2 and Gemini 3.0 Pro on some tasks. In coding competition benchmarks, DeepSeek said both V4 models' performance is "comparable to GPT-5.4."

Notably, the flagship V4-Pro model features a total of 1.6 trillion parameters using a mixture-of-experts architecture - meaning only a fraction are active at any one time - making it the largest open-source model available. (Open-source means that it can be downloaded and used privately without being monitored, removing any safeguarding implemented by the original developers). It is also much more affordable than any frontier model available today.

These developments have escalated alarms among the AI safety community. The most powerful AI tools are becoming cheaper and more accessible faster than anyone expected. And the powerful ones might have already slipped out.

TL;DR: In April, Anthropic unveiled Mythos - an AI model capable of autonomously hacking into any major operating system - and restricted it to a broad coalition of trusted partners for safety. Days later, DeepSeek released an open-source model nearly matching the world's best proprietary AI. The gap between cutting-edge and freely available is now measured in months.

ASTROBIOLOGY

New Building Blocks of Life on Mars Discovered

In October 2020, a car-sized NASA rover named Curiosity drilled into a slab of clay-bearing sandstone within the Gale Crater on Mars. The specific site being drilled was named after Mary Anning, the 19th-century fossil hunter who spent her life finding evidence of ancient life.

That naming turned out to be apt. Recently published in Nature Communications, Curiosity’s findings reveal 21 carbon-containing organic molecules in that single sample, the most diverse collection of building blocks of life on Mars. Seven had never been detected on the planet before.

The breakthrough came from a new technique. Instead of the usual approach of baking rock powder in an oven - which destroys the most delicate molecules before they can be identified - the team dissolved the sample in tetramethylammonium hydroxide, a solvent that gently breaks apart organic compounds.

Among the finds was a nitrogen heterocycle: a ring of carbon atoms containing nitrogen, considered a direct precursor to DNA and RNA. "Nitrogen heterocycles have never been found before on the Martian surface or confirmed in Martian meteorites," said lead author Amy Williams of the University of Florida.

Scientists have no way of knowing whether these molecules were created by biological or geological processes. Either path is possible. But that ambiguity is itself revealing: ancient Mars had the right chemistry for life to begin, and that chemistry survived 3.5 billion years of radiation. The molecules are preserved. The question of what made them remains open. ​

TL;DR: NASA’s Curiosity rover has uncovered the most diverse array of organic molecules ever found on Mars, including seven that had never been detected before on the red planet. Whether these molecules were created by biological or geological processes is unknown.

SPACE

NASA Shuts Off Instrument on Voyager 1
to Keep Spacecraft Operating

In February, during a routine roll manoeuvre, Voyager 1's power levels dropped. Engineers knew that any further decline could trigger the spacecraft's undervoltage fault protection system - an automatic shutdown that carries its own risks and requires a lengthy recovery. The team needed to act first.

On April 17, they shut down the Low-Energy Charged Particles (LECP) experiment, an instrument on the Voyager that had run almost without interruption since Voyager 1 launched in 1977. Only two of its original 10 science instruments now remain powered - one that listens to plasma waves, one that measures magnetic fields.

Voyager 1 was designed for a five-year mission. It is now 16 billion miles from Earth - the most distant object humanity has ever built - and still transmitting data. But perhaps not for much longer. Its nuclear power supply loses about 4 watts per year, and power margins have become razor-thin.

This shutdown buys about a year. The team is using that time to prepare something more drastic: a manoeuvre they call "the Big Bang." The plan involves turning off a group of powered devices simultaneously and activating low-power alternatives to keep the spacecraft warm enough to continue gathering data. Voyager 2 will serve as the test subject in May and June. If successful, the same manoeuvre follows on Voyager 1 in July. And if that works, the LECP could be reactivated.

TL;DR: After an unexpected power drop in February, NASA shut down Voyager 1's Low-Energy Charged Particles instrument, leaving the 49-year-old spacecraft with just two working sensors as it travels 16 billion miles from Earth. A bold "Big Bang" power swap planned for July 2026 is the mission's best shot at survival beyond this year.

News

Robots are finally beating elite table-tennis players. A robot called Ace just won 3 of 5 matches against elite players using AI trained entirely in simulation, with superhuman reaction speed and precision that even surprised Olympic-level opponents. It's the first time a robot has outplayed humans at the sport.

MIT found a bridge between quantum and classical physics. For over a century, quantum mechanics and classical physics have been treated as two separate rulebooks (one for subatomic particles, one for everything else). MIT researchers just found they may be the same book. By extending a classical framework to incorporate the concept of path density, they reproduced the exact results of quantum mechanics, including famously weird phenomena like particles tunnelling through walls. In the case of the classic double-slit experiment, this can be reduced to just two classical paths rather than the infinite tangle quantum theory requires.

We finally know how melatonin actually puts you to sleep. Millions of people take melatonin supplements to sleep, but the exact neural mechanism was unknown. Until now. Caltech researchers have found the answer in zebrafish, which share sleep-regulating biology with humans, and it's surprisingly intuitive. Melatonin promotes sleep by binding to receptors in the brain's visual processing centre, quieting neurons that respond to light and making the brain less reactive to visual stimuli.

A new measure of gravity. We've been trying to measure the strength of gravity for over 225 years and we still can't agree on the answer. NIST physicists have arrived at a new value for G: 6.67387×10⁻¹¹. The result is at odds with both the French measurement it was designed to replicate and the broader international consensus value, meaning the persistent disagreement between measurements remains unresolved. But by identifying where previous experiments may have gone wrong, this decade-long effort brings new rigour to one of physics' most stubborn puzzles.

Quantum-safe encryption for tiny devices. MIT engineers have built a microchip the size of a needle tip that brings quantum-resistant encryption to wireless biomedical devices like pacemakers and insulin pumps, hardware normally too power-constrained to run such demanding security. The new chip is more than 10 times more energy-efficient than existing approaches, and includes built-in defences against physical hacking.

Could your morning coffee insulate your home? Every year, millions of tons of used coffee grounds go straight to landfill. Researchers have found a better destination: your walls. By baking spent grounds into a highly porous material and combining it with ethyl cellulose, a natural polymer, they created an insulator that performs six times better than the polymer alone, matching the thermal performance of commercial expanded polystyrene. Unlike polystyrene, it's biodegradable, made entirely from waste, and requires no petroleum.

Astronomers have just seen the universe’s very first stars for the first time. Using the James Webb Space Telescope, astronomers have detected the most compelling evidence yet for the universe's first generation of stars. A tiny gas cloud near one of the earliest known galaxies emits only hydrogen and helium signatures with no heavy elements at all (the chemical fingerprint theorists have long predicted for these primordial stars). The object, nicknamed Hebe, sits just 400 million years after the Big Bang. Two independent teams have now confirmed the signal.

This Week in History

April 30, 1897. In a lecture that became a milestone in physics, J. J. Thomson announced he had found something smaller than the atom, a “corpuscle”, later called the electron. At the time, atoms were thought to be indivisible, so the claim seemed almost absurd. But this tiny particle became the first known building block of matter, opening the door to modern atomic and particle physics.

April 26, 1920. Mathematician Srinivasa Ramanujan died at just 32, leaving behind mathematics that feels like it arrived as divine inspiration. Largely self-taught, he uncovered deep results in number theory, infinite series, and partitions, often without formal proofs, yet later confirmed to be true. His intuition was so sharp that when his mentor, G. H. Hardy, pointed out how mundane their taxi number 1729 was, he could instantly respond that this number was actually quite remarkable: it is the smallest integer that can be represented in two ways by the sum of two cubes.

April 26, 1986. A late-night experiment at the Chernobyl nuclear plant spiraled into the worst civilian nuclear disaster in history. Reactor 4 exploded, releasing a radioactive cloud across Europe and exposing millions, with immediate deaths followed by long-term health consequences still unfolding decades later.

This Week’s Puzzle

🧩

CATCHING A RIDE WITH 3 SCIENTISTS

Three scientists - Einstein, Feynman, and Curie - each offer you a ride. Two want to kill you. One doesn't.

Rules:

• Exactly one scientist is safe to leave with
• At least one scientist always lies (could be safe or dangerous)
• All three statements below are given to you as fact

Their statements:

• Einstein: "Feynman and Curie both tell the truth."
• Feynman: "To survive, choose Einstein or choose Curie."
• Curie: "Feynman is NOT the one to choose if you want to live."

Question: Which scientist should you choose to leave with?

Until next time,

The Ve Team 👋

Know someone who would enjoy this newsletter? Forward it their way.

Research by Pablo
Written by Ines

​Unsubscribe · Preferences​

🧩

​
Solution

The answer is Feynman.

Most people focus on Feynman's and Curie’s statements. They seem to be the "content-rich" clues. But the real key is Einstein's statement, which is easy to skim past.

Einstein's statement is self-defeating. If he's telling the truth, then Feynman and Curie also tell the truth, meaning all three are truthful. But we know at least one always lies. So Einstein must be lying.

Since Einstein is lying, it’s not the case that both Feynman and Curie tell the truth. At least one of them lies.

We can not test the remaining cases:

Case A: Both Feynman and Curie lie. Feynman says "choose Einstein or Curie". If this is a lie, neither is safe, so Feynman is safe. Curie says "don't choose Feynman". If this is a lie, Feynman is safe. Both point to Feynman. There’s no contradiction here.

Case B: Feynman lies, Curie tells truth. Curie says Feynman is not safe → Feynman is dangerous. Feynman lies saying "choose Einstein or Curie" → both are dangerous. That makes all three dangerous, which is a contradiction.

Case C: Feynman tells truth, Curie lies. Feynman says "choose Einstein or Curie" → one of them is safe. Curie lies saying "don't choose Feynman" → Feynman is safe. But Feynman just said Einstein or Curie is safe. Two safe scientists is impossible, so this is a contradiction.

So only Case A holds and you should leave with Feynman.