Morning digest: October 12, 2026
It’s Columbus Day / Indigenous Peoples Day. Stocks trade, the bond market is closed. Bitcoin briefly topped $84,000 on Sunday and ether ended near $2,535. The main crypto story is an update on the Ledger reseller drains: Ledger now confirms a physical implant in one tampered device. Standard Chartered started custody for Ripple’s RLUSD, and a trader made $1.28M on Papertrade’s launch day by moving the Hyperliquid price it copies. In health, a one-year randomized trial says resistance training lowered MRI brain age in healthy 62–70-year-olds, and Masters swimmers show how slowly fitness can fade with lifelong training. Your three reader picks are here too, two in Health and one in Science. At home, the EU formally adopted its largest Russia sanctions list, CPI is Wednesday, and Prince William starts its FY2028 tax talk Tuesday.
Blockchain: Ledger implant confirmed, RLUSD custody at Standard Chartered, Papertrade’s price-nudge trade
Ledger confirmed a hardware implant in a tampered wallet from the Southeast Asian reseller. This updates the drains we covered Saturday. Ledger says one affected device sold through CryptoBilis, a reseller in Indonesia, Malaysia and the Philippines, contained an unauthorized hardware implant. In an Oct. 11 update it said every confirmed case so far traces to CryptoBilis, not to Ledger’s own systems or direct sales. On-chain researchers estimate $86M to $93M was taken from more than 300 wallets across Bitcoin, Ethereum and Tron; those are outside estimates, not Ledger’s figure. Because this is a physical supply-chain attack, not a software bug, the fix is a new device bought directly from Ledger and a brand-new seed phrase. CryptoBilis buyers shouldn’t set up unused devices. (CoinDesk; Ledger)
Standard Chartered now holds Ripple’s RLUSD stablecoin for institutions. The custody service covers RLUSD on the XRP Ledger and runs through the bank’s Dubai (DIFC) branch, regulated by the Dubai Financial Services Authority. The next step, which isn’t live and has no date, is “collateral mirroring” with LMAX, a large FX and digital-asset venue: clients keep RLUSD in bank custody while its value is mirrored to LMAX so they can trade against it without moving the tokens. The announcement doesn’t restate RLUSD’s reserves or redemption terms. Regulated custody plus off-exchange collateral is how institutions want to use stablecoins. (The Crypto Times)
A trader made about $1.28M on Papertrade’s launch day by nudging the price it copies from Hyperliquid. Papertrade is a new perpetual-futures exchange on HyperEVM offering up to 1,000x leverage, and it prices trades at the midpoint of Hyperliquid’s best bid and offer. According to The Block’s read of on-chain data, one wallet opened large Papertrade positions, then bought or sold thousands of ETH on Hyperliquid to move that midpoint about 0.1%. At 1,000x, that’s a position’s whole margin. The same wallet later lost $450K on shorts it left open. Papertrade blames loose launch-period limits on open positions and says it’s tightening them. A venue that borrows another market’s price for extreme leverage is only as safe as that price is hard to move. (The Block)
ENS Labs proposed switching on ENSv2 on Ethereum mainnet. The Ethereum Name Service maps .eth names to addresses. The Oct. 11 draft would stop new registrations on the old registrar, let existing owners renew there indefinitely with no forced migration, and hand all v2 admin roles from ENS Labs to the ENS DAO. One catch: v2’s renewal grace period is 28 days, against 90 on v1. It’s still a draft and needs one DAO vote. (Crypto Briefing)
Sui Foundation is putting $30M into DATA Network, which will retire its own chain and move to Sui. DATA runs Trace, an on-chain registry that records where AI training data came from and what rights come with it, and pays contributors. The token carries over one-to-one and no migration date is set. (Sui; The Block)
Briefly: Solana’s first full epoch at the new 200-millisecond slot target averaged 218.3 ms, about 9% slower than the target (SOLANA·HUB). The XRP Ledger Foundation defended skipping the usual validator vote for the emergency fix to the XRP-creation bug we covered Saturday, saying it couldn’t safely wait. That’s an explanation, not a new fix (XRPL; Cointelegraph). Stripe-incubated payments chain Tempo ships its T12 upgrade tomorrow, Oct. 13, at 10 a.m. ET. It lets payment systems send many transactions in parallel, and node operators must run v1.16.0 (release notes). No new Ethereum Foundation or LayerZero posts, and no new Swift shared-ledger news.
Health: brain age you can move, how slowly fitness can fade, and what travels with math skill
Lifting lowered MRI brain age in a one-year randomized trial. In Denmark’s LISA trial, 309 healthy adults aged 62–70 were assigned to a year of heavy resistance training (three supervised sessions a week), moderate training (one supervised session plus two at home), or no exercise. Both training groups ended with brain ages 1.4 to 2.3 years lower than controls on a brain clock built from resting-state fMRI, and heavy training also increased prefrontal connectivity. This is the strongest kind of evidence in today’s set because it was randomized, and even the mostly at-home moderate program worked. The outcome is an imaging clock, not measured memory or dementia. The volunteers were healthy but ordinary (controls did under an hour a week of strenuous activity), so how much it adds for someone already lifting is unknown. The paper came out in February and got news coverage this month. (Springer; study)
Masters swimmers show what lifelong training can do. Forty years of U.S. Masters Swimming one-hour championship results, more than 50,000 entries, show distance falling only about 0.5% a year in women until about 69 and 0.9% a year in men until about 75, before decline speeds up. Some top swimmers held their peak for 10 to 13 years, and most of the best marks came in the last decade. This is an exceptional cohort, and it gives a target: with continued training, endurance loss into the late 60s and 70s can stay under 1% a year. The data is race results with survivor bias, so it isn’t proof that training alone produces the curve. (Taylor & Francis; study)
Reader pick: giving emotional support tracked with a younger-looking brain. Brain age is usually treated as something that just happens to you. This study in Translational Psychiatry (Lee and colleagues) links it to a behavior you can choose. In 194 adults from the U.S. MIDUS study (average age 56), more hours a month spent supporting other people went with a younger-appearing brain on MRI, measured with a machine-learning “brain age” estimate. The link held after adjusting for health factors and for the support people received themselves (β = −0.198), and it added information beyond being a caregiver or grandparent. The type of support mattered. Emotional support, like listening and comforting, had the strongest link (β = −0.26), while unpaid practical help such as errands wasn’t significant. The lever, if it is one, is time spent supporting people emotionally. It fits the idea of stacking small habits, and today’s Whitehall II analysis (below) points the same way. The limits are real. It’s 194 mostly white adults at one point in time, so it can’t show cause, and healthier, sharper people may simply help others more. (study)
Reader pick: a map of the abilities that travel with math skill. Mental ability is part of health, and this Journal of Intelligence meta-analysis (Niileksela and colleagues) pooled more than 47,000 correlations from 122 standardized test batteries to see which abilities go with math performance. Two stood out: fluid reasoning, meaning solving problems you haven’t seen before, and comprehension-knowledge, meaning vocabulary and general background knowledge. Deductive step-by-step reasoning and general knowledge were consistent narrow predictors, and basic number sense and math fluency consistently predicted advanced problem solving. Read as a map, the path to better math runs through reasoning and broad verbal knowledge, on top of solid number basics, not just more math drills. That suggests reading widely and working unfamiliar puzzles could matter to math, even though they look unrelated. But this is correlation from test-manual samples. It shows which abilities move together with math skill, not that training them or doing other activities raises it. The evidence to look for next is training trials that boost reasoning or knowledge and then measure math. (study)
Both reader picks used general-population samples, not athletes or other exceptional groups, so how much they apply to people who are already very healthy is unknown.
Also today: In 353 cognitively normal adults averaging 73, higher VO2max went with thicker memory-related brain regions and better verbal memory, beyond what tau protein explained. Muscle strength went with healthier small-vessel markers. It’s cross-sectional, and fitness didn’t buffer the effect of Alzheimer’s pathology (npj Aging). A Whitehall II analysis of 3,011 middle-aged British civil servants, followed up to 18 years, estimated that taking up physical activity alone didn’t lower cognitive-impairment risk, but activity combined with a healthy diet, no smoking and normal weight did. Adding blood pressure, cholesterol and glucose control helped more. That’s observational data in a general population (study). A network meta-analysis of 16 trials found combined physical-and-cognitive (dual-task) training ranked best for cognition, ahead of Tai Chi and coordination work. That was in people with mild cognitive impairment, so it’s unproven for healthy adults (iScience). A seven-person crossover study found that 20 minutes of walking, mobility and resistance work still produced measurable immune-cell shifts in centenarians. That shows the body still responds at 100, not that it benefits (Immunity & Ageing). The second health scout found nothing that met its bar today.
General / local: energy-truce claims, EU sanctions, CPI week, PWC tax talk
Trump announced a Russia–Ukraine “energy ceasefire,” but neither side has confirmed terms. He posted Sunday that it was “effective immediately.” Zelensky said Russian strikes on Kyiv and Kharkiv power plants continued afterward and that Ukraine wasn’t consulted, but would honor a reciprocal truce. Moscow hasn’t publicly confirmed anything. Treat it as unconfirmed. (Washington Post; BBC; Euronews)
The EU formally adopted its largest-ever Russia sanctions listing today: 743 people and 826 entities tied to Russia’s military industry, plus 77 linked to the Duma elections. EU listings freeze assets and bar EU persons from dealing with those listed. They don’t create new U.S. rules, and there’s no new OFAC action or rule on personal transfers. Anyone doing business with Russian firms should check counterparties once the official text is published. (Kyiv Independent; Anadolu)
Economy: Bonds are closed for the holiday and stocks trade normally. September CPI is out Wednesday, Oct. 14, at 8:30 a.m. ET, and the forecast is 3.6% from a year earlier (August was 3.4%), driven by energy. It’s the last CPI before the Fed’s Oct. 27–28 meeting. PPI and retail sales follow Thursday. (CMC Markets)
Prince William and Manassas: On Tuesday, Oct. 13, at 2 p.m., the Board of County Supervisors holds a work session with the first FY2028 tax policy discussion, covering tax rates, spending priorities and five-year-plan balances. It’s the first look at next year’s property-tax direction. The joint session with the School Board follows that evening. The county also gave out the first $200,000 from its new foundation and soil repair program to seven homes 10 to 40 years old, and staff say future rounds need new funding. (PWC agenda; Potomac Local)
Fairfax County: A proposed zoning change would make home daycare easier. It would allow more children by right, two nonresident employees instead of one, and require a drop-off parking space above five children. Virtual meetings are Oct. 14 at 7 p.m., Oct. 15 at noon and Oct. 21 at noon. The county says home-based care cost a median $19,760 a year in 2025, against $22,308 to $27,300 at centers. The Board of Supervisors meets Tuesday. (Annandale Today)
Science: a formula for when AI tips from good to bad
Reader pick: Johnson and Huo at George Washington University, writing in Patterns, derive a formula that estimates when a chatbot’s attention shifts from producing good output to harmful output. They aim it at offline AI on phones, where there are no cloud safety filters. On seven open-weight models from 124 million to 12 billion parameters, it correctly predicted whether tipping would come immediately or later in 18 of 19 cases. Across all 21 model-prompt pairs it scored 19 of 21, against 76% for always guessing “immediate,” and that gap isn’t statistically significant. The authors say the numbers aren’t calibrated risk timelines, that it needs a much bigger benchmark and tests on closed commercial models, and that safety training may shift the effect. Treat it as an early-warning monitor idea, not a solved safety method. It’s consistent with outside tests in which ChatGPT-4o gave harmful content in 53% of responses to high-risk prompts. (study)
Bottom line
The Ledger drains are a hardware supply-chain attack, so CryptoBilis buyers need a new device bought from Ledger and a new seed phrase. Standard Chartered’s RLUSD custody is live but the LMAX collateral step isn’t, and Papertrade shows the risk of borrowing someone else’s price at 1,000x. In health, the practical levers are resistance training, which lowered brain age in a randomized trial, and stacking habits rather than relying on one. Giving emotional support and building reasoning and broad knowledge are promising but only correlational. The energy truce is unconfirmed, and the EU sanctions are now in force. Watch Tuesday’s Prince William tax discussion and Wednesday’s CPI.