BTC $83,127 +1.4%ETH $2,496 +1.5%Fear & Greed 57 Greed

Today at a glance

A hardware-wallet supply-chain implant and Solana's consensus overhaul land together, forcing a joint reassessment of crypto infrastructure security and speed.

3 signals
  • Wallet supply chainLedger confirmed a spy implant in CryptoBilis-sold devices that decodes recovery phrases, and Trezor responded by ending its authorized reseller network on January 1, 2027.#01
  • Solana upgradeAnza said Alpenglow activates on mainnet on October 21, replacing TowerBFT and cutting finality from seconds to about 150 milliseconds.#02
  • AI frontierGoogle DeepMind published pre-release safety documentation for Gemini 4 Argon and announced a robotics version, without disclosing benchmarks or a general-availability date.#05

Stories are ranked by impact; the first three are the edition highlights. This edition displays 14 of 592 candidates.

#01
CryptoEdition highlightTracking · 42 updates
8.5

Trezor Cuts Its Resellers After Ledger Finds Spy Implant in Crypto Wallets

Background

Background

CryptoBilis is listed as an official Ledger reseller in Indonesia, Malaysia and the Philippines, and Ledger asked the company to pause all sales and shipments during the investigation. In May 2022, Trezor warned that counterfeit Model T wallets were being sold on a Russian marketplace by a non-authorized seller, with opened, resealed devices whose recovery phrases were predictable or predefined; its advice then was to buy only from its official shop or an authorized reseller. Trezor also said earlier this week that CryptoBilis was already off its seller list and that it had received no reports of tampered Trezor devices from that retailer.

Trezor has told partners in its authorized reseller network that their authorization ends on January 1, 2027, and that it will ship them no new stock, days after rival Ledger's Donjon security team confirmed a hidden spy implant in hardware wallets sold by Southeast Asian reseller CryptoBilis. Trezor said there is "not a single piece of evidence" its own devices were compromised and described the cut as a precaution.

The Ledger finding is a confirmed supply-chain compromise of seed-phrase handling in widely used hardware wallets, and Trezor's decision shows the incident is already reshaping how wallet makers manage third-party distribution.

Ledger says the CryptoBilis implant reads the screen and decodes the recovery phrase when it appears, then sends it out over LTE via an eSIM, without altering the key-storage chip or the operating system, so a tampered wallet looks and works normally. Onchain investigator Specter traced more than $86 million in suspected thefts, a figure Ledger says it has not verified; Trezor has not said how many resellers are affected, and those cut can keep selling existing stock through the holiday season.

rss · Decrypt · · 2 sources

Analysis and references

Impact analysis

The incident transmits mainly through hardware-wallet distribution and self-custody trust: buyers of devices through third-party resellers face elevated counterparty risk, and wallet makers' reseller vetting, device sourcing and after-sale verification become the exposed segment rather than any specific token. Ledger's paused CryptoBilis sales and Trezor's shrinking reseller channel both shift volume toward direct-from-vendor purchases.

References

Tags

#hardware-wallets#supply-chain-attack#security-incident#ledger#trezor#cryptobilis

#02
CryptoEdition highlightTracking · 8 updates
8.5

Solana consensus upgrade Alpenglow to go live on mainnet on October 21, cutting finality to about 150 milliseconds

Background

Background

TowerBFT has been Solana's consensus mechanism since mainnet launch, working alongside Proof of History (PoH), a cryptographic clock that orders events. Anza and Solana describe Alpenglow as the most significant overhaul of Solana's core protocol to date; the new voting protocol Votor ships in Agave 4.3.

Anza, Solana's core development team, announced that the Alpenglow consensus upgrade is scheduled to activate on Solana mainnet-beta around October 21 at 22:50 (UTC+8, roughly 14:50 UTC), with the migration to be completed within a single slot. Alpenglow replaces TowerBFT, the consensus protocol Solana has used since genesis, and is intended to cut transaction finality from seconds to about 150 milliseconds.

Anza says moving voting off-chain frees block space for user transactions, and that faster finality shortens settlement for institutions from the usual 1–3 business days to roughly 150 milliseconds. It is described as the most significant change to Solana's core consensus since the network launched.

Anza developed Alpenglow over 18 months; it was previously migrated on testnet (September 24) and devnet (September 25) and tested in a 1,500-node simulation, with a bug bounty exceeding 25,000 SOL. Under the new Votor voting protocol, validators exchange votes directly instead of submitting them as on-chain transactions, which the announcement says could cause a significant decline in reported transaction totals that include validator votes.

telegram · foresightnews · · 4 sources

Analysis and references

Impact analysis

The upgrade directly concerns SOL stakers and validators and the applications settling on Solana, since sub-second finality could change how exchanges, custodians and payment processors treat on-chain settlement. Because validator votes move off-chain, widely cited Solana transaction-count metrics may shift, affecting how data providers and market observers read network activity.

References

Tags

#solana#alpenglow#towerbft#votor#consensus-upgrade#mainnet-launch

#03
AI & TechEdition highlightTracking · 2 updates
8.5

Just 5 years in, interactive chatbot platform Character.AI files for bankruptcy

Background

Background

Character.AI was founded by Noam Shazeer, a co-author of the Transformer paper, and former Google researcher Daniel De Freitas; its app drew about 1.7 million installs within a week of its 2023 launch, and the company raised $150 million at a $1 billion valuation. In 2024, Google paid $2.7 billion for a non-exclusive license to Character.AI's model technology and hired the two founders plus roughly 20% of staff, after which Character.AI dropped frontier-model competition to focus on its chat product. The company later faced lawsuits over harm to minors, and search results indicate it agreed with Google in January 2026 to settle five such suits.

Character.AI filed for Chapter 11 bankruptcy protection in the United States on October 12, roughly five years after launch, and Disney has submitted a $20 million initial bid for some of its assets. The company expects to complete the sale between late November and early December.

The platform has not shut down yet, but the company says it will stop service if no buyer is found to keep operating it, and users have already begun asking how to preserve chat histories with their AI characters that span several years.

Disney's initial offer covers only a perpetual, non-exclusive license to some of Character.AI's intellectual property and related rights, and the entertainment company has not committed to taking over the whole platform. Character.AI's former CEO Karandeep Anand joined Disney as chief technology officer on October 2, with some technical staff moving with him; those departed employees are helping keep Character.AI running temporarily during the sale process.

rss · TechCrunch AI · · 2 sources

Analysis and references

References

Tags

#Character.AI#Chapter 11 bankruptcy#Disney#AI startups#consumer AI#minors safety litigation

#04
Crypto
8.0

XRPN took shareholder vote with no knowledge of XRP inflation bug

Background

Background

XRP is a pre-mined digital asset whose full 100 billion supply was created at the XRP Ledger's launch in 2012, and the protocol has no mining or staking mechanism for issuing new tokens. Evernorth Holdings, an XRP treasury company backed by Ripple and Arrington Capital, completed its business combination with Armada Acquisition Corp. and, per search results, is capitalized with roughly 473 million XRP and $300 million in gross cash proceeds.

A newly disclosed bug in the XRP Ledger's base-layer payment engine could have allowed an attacker to mint trillions of new XRP for almost nothing; researcher Cayden Liao and Veria AI reported it on September 22, and Ripple shipped a closed-source fix in xrpld 3.4.1 on September 25 described only as "integer-arithmetic hardening." Shareholders of the Ripple/Arrington-backed XRP treasury vehicle Evernorth (XRPN) voted on their plan without being told about the flaw, and an official report admitting what the patch actually fixed appeared only on October 9, the day the merger closed.

The flaw meant XRP's fixed supply of 100 billion — the core claim underpinning Evernorth's treasury plan — was technically unenforceable on the network for roughly a decade. The episode also raises governance questions, since XRPN shareholders made redemption and merger decisions inside a window in which Ripple knew about the bug but they did not.

According to the report, the bug likely dated back to 2015 and a single validated transaction using the exploit "could have created spendable XRP far beyond the total supply." The patch shipped without an amendment vote, and no actual exploitation of the vulnerability has been reported.

rss · Protos · · Single source

Analysis and references

Impact analysis

The disclosure bears directly on the credibility of XRP's supply cap, a claim that underpins the valuation of the token itself and of treasury vehicles such as Evernorth/XRPN whose equity is tied to XRP holdings. It also draws scrutiny to XRPL validator and patch-governance practice, including the emergency closed-source release issued without an amendment vote.

References

Tags

#XRP Ledger#protocol vulnerability#Evernorth/XRPN#token supply#governance disclosure

#05
8.0

Developing Gemini 4 Argon responsibly and safely

Background

Background

Frontier labs have increasingly published safety and evaluation documentation alongside new model announcements, a practice that search excerpts link to pressure from regulators and safety bodies such as NIST and the UK AI Safety Institute for mandatory frontier evaluations. Google DeepMind's post follows that format rather than reporting a capability result or a safety incident.

Google DeepMind published a blog post outlining how it is developing its frontier model Gemini 4 Argon responsibly and safely, detailing the evaluations and mitigations applied ahead of its release.

The post places Gemini 4 Argon at the stage of public pre-release safety documentation, adding it to the small group of frontier models whose deployment preparations are described in the open.

A related post on Google's The Keyword announcing the model says Gemini 4 Argon is designed to sustain deep reasoning across complex, long-horizon workflows and is rolling out to a set of trusted cyber defenders through Google's Fairwind Program. Neither post, as reflected in the available material, discloses benchmark figures or a general-availability date.

rss · Google DeepMind Blog · · Single source

Analysis and references

Impact analysis

The post is process- and safety-oriented and does not itself change any crypto-specific rule, custody arrangement or liquidity condition. Any transmission runs through sentiment toward AI-themed and decentralized-compute tokens, which frequently react to frontier-model news from major labs.

References

Tags

#AI safety#Gemini 4 Argon#frontier models#Google DeepMind#model deployment#Fairwind Program

#06
AI & Tech
8.0

Gemini 4 Argon for Robotics

Background

Background

Embodied AI refers to AI systems embedded in a physical body that perceive their environment through sensors, act through actuators and pursue goals over time, a shift from purely digital AI into the physical world. Foundation models for robotics are being applied in this direction, with commentators describing the current period as robotics' 'GPT-2 moment' while cautioning that model scale alone does not make a robot work in practice.

Google DeepMind announced Gemini 4 Argon for Robotics, a new generation of its Gemini model family adapted for robotics and embodied AI applications.

It extends Google DeepMind's flagship Gemini family from software-facing tasks into embodied AI, the part of the field concerned with systems that perceive and act in the physical world through sensors and actuators.

Coverage of the wider Gemini 4 Argon release states that it was presented on 30 September 2026 with a single-trajectory output capacity of 1 million tokens and priority access for cyber defenders — features described for long-horizon enterprise and software-development work rather than robotics specifically.

rss · Google DeepMind Blog · · Single source

Analysis and references

Impact analysis

Coverage of the broader Gemini 4 Argon launch linked it to a dip in Alphabet shares while the wider tech sector rose, indicating the near-term transmission runs through equity sentiment toward large AI platform owners rather than through crypto fundamentals. Crypto exposure is indirect, via the general AI-narrative and risk-appetite channel that links mega-cap tech moves to digital-asset sentiment.

References

Tags

#ai#google-deepmind#robotics#foundation-models#embodied-ai

#07
AI & Tech
8.0

Higgsfield takes AI video from 1 minute to over 10

Background

Background

OpenAI's GPT-6 Astra was initially released to approved users on September 3, 2026, with general availability the following day, and is offered in the OpenAI API as gpt-6-astra as well as through Microsoft Azure and Amazon Bedrock, with listed standard pricing of $10 per million input tokens and $50 per million output tokens. OpenAI has also introduced ChatGPT Images 2.5, branded 'Sunburst', aimed at premium visual and editing workflows. Higgsfield describes itself as an AI-native creative suite that generates images, videos and voice content from prompts or references on web and mobile.

OpenAI says its GPT-6 'Astra', GPT-Image-2.5 and GPT-Live models are now powering Higgsfield, allowing creators on the platform to generate videos longer than 10 minutes with more precise visuals and voice-driven filmmaking.

According to the announcement, the integration extends AI video generation on Higgsfield from roughly one minute to over 10 minutes, a change aimed at longer-form creative work rather than short clips.

The announcement is framed as a partner story and names three models: GPT-6 Astra for longer videos, GPT-Image-2.5 for precise visuals, and GPT-Live for voice filmmaking. Higgsfield's platform separately aggregates third-party video models such as Kling 3.0, Veo 3.1 and Sora 2.

rss · OpenAI Blog · · Single source

Analysis and references

Impact analysis

The story has no direct on-chain, custody or protocol exposure; any transmission runs through sentiment toward AI-related crypto narratives and through continued demand for the GPU and API capacity that long-form video generation consumes.

References

Tags

#OpenAI#GPT-6 Astra#GPT-Image-2.5#Higgsfield#video-generation#multimodal

#08
8.0

Anthropic warns AI worms are closing in: 15 cheap TV boxes can already run a 26-billion-parameter model

Background

Background

AI-worm research has been building for some time: IBM described the Morris II proof-of-concept worm that targets generative-AI email assistants, and security vendors classify AI worms as malware that injects self-replicating prompts rather than dropping executable files. In June 2026, University of Toronto researchers reported building and testing a proof-of-concept AI-driven worm that used a locally hosted open-weight large language model to reason its way through a network and replicate itself.

Anthropic published research warning that AI could one day spread like a computer worm — autonomously finding networked devices, deploying models on them and using those devices to propagate further, without necessarily needing data centers. In its tests, 15 Android TV boxes priced at about $55 each ran a split 26-billion-parameter Gemma 4 model at only 0.39 tokens per second, Alibaba's Qwen3.5-9B autonomously stood up inference services on 8 of 9 hardware types, and Zhipu's GLM-5.2 completed a find-transfer-run-propagate loop in an isolated simulated network, covering all 8 target nodes in 4 of 5 runs and 7 of 8 in the fifth.

The research puts a concrete hardware floor under autonomous AI propagation, showing that ordinary consumer devices can serve as a compute source, and Anthropic projects that small models suited to cheap devices may reach GLM-5.2's current simulated-propagation capability by the end of 2026.

The demonstrations are bounded: the experiment deliberately opened an unauthenticated debug interface, which greatly reduced the difficulty of intrusion, and GLM-5.2 itself ran on external servers — what was deployed to target devices was the smaller Llama 3.1 8B, so it did not achieve true self-replication independent of external compute. Based on current experimental data, a 9B model might need one to two days to complete one round of replication.

telegram · theblockbeats · · Single source

Analysis and references

Impact analysis

The research does not directly touch any protocol or token; any market transmission would run through sentiment rather than fundamentals, as AI-safety and self-propagating-AI headlines can shape the narrative around AI-themed crypto assets and decentralized edge-compute projects without changing their cash flows or usage.

References

Tags

#ai-safety#self-replicating-ai#anthropic#edge-inference#ai-security

#09
Crypto
7.5

BNY, Citadel Securities test tokenized U.S. Treasuries to speed up Wall Street financing

Background

Background

Wall Street has been exploring bringing traditional assets onto blockchains, with firms including BlackRock, Franklin Templeton and JPMorgan embracing tokenized funds and securities on bets the technology can shorten settlement times, automate processes and make assets easier to move. BNY has separately moved into digital asset infrastructure, including custody and issuance support for the USDC stablecoin, and has said it plans to extend support to other stablecoin issuers and digital cash services. Tokenized U.S. Treasuries are blockchain-based representations of government debt such as short-term Treasury bills.

BNY, described as the world's largest custody bank, and trading firm Citadel Securities said on Tuesday they had completed a transaction using tokenized U.S. Treasuries, showing how blockchain-based government bonds could move from settlement into financing. BNY also said it plans to add digital cash and tokenized deposit capabilities and to extend the operating hours of its traditional settlement systems for Federal Reserve-eligible securities and U.S. triparty transactions, with the aim of reaching European and Asian trading hours next year.

The effort targets a key bottleneck in financial markets: moving government bonds used as collateral for trading and borrowing can take time, tying up assets that could otherwise be put to work. Citadel Securities digital assets operations lead Dan Doherty said efficient movement of high-quality collateral is fundamental to well-functioning markets and that reducing delays across the trade lifecycle could improve how institutional markets operate.

The transaction was a demonstration of how tokenized government bonds could move from settlement into financing, potentially letting firms reuse collateral sooner and free up cash for other trades. BNY chief innovation and market transformation officer Carolyn Weinberg said using a tokenized representation of a U.S. Treasury across both settlement and financing enables a more connected, efficient and always-on financial system.

rss · CoinDesk · · Single source

Analysis and references

Impact analysis

The transaction sits in the tokenized real-world asset and collateral segment: if tokenized Treasuries can move between settlement and financing rails, they could touch repo and triparty collateral flows where USDC and other digital cash instruments are used for settlement, a channel that large custodians and stablecoin issuers are already building toward. The described plans remain pilot-stage and depend on BNY extending digital cash services and settlement hours.

References

Tags

#tokenization#tokenized-treasuries#BNY#Citadel Securities#collateral#market-structure

#10
Crypto
7.5

Ethereum Name Service plans major .eth overhaul with one irreversible catch for domain holders

Background

Background

ENS is a naming system on Ethereum that maps human-readable names such as .eth names to wallet addresses and other identifiers. ENSv2 was first outlined in 2024 as a ground-up rebuild of the protocol, and ENS's FAQ states that name ownership is not affected by the transition. The ENSv2 page offers existing .eth name holders an upgrade path for their names.

Ethereum Name Service is planning a major overhaul of .eth domains that, according to CryptoSlate, includes an irreversible catch for domain holders. Coverage of the proposal points to an ENSv2 upgrade featuring a one-way domain migration option.

The change targets .eth, ENS's core naming product, and the described one-way migration gives holders a decision that reportedly cannot be reversed once taken. ENS's own FAQ states that existing name ownership would be preserved through the ENSv2 transition.

Coverage of the proposal cites an irreversible, one-way domain migration, support for stablecoin payments, and a 60% fee increase for five-character names. ENS's ENSv2 page describes a protocol rebuilt from the ground up with a new ownership model and new apps.

google_news · CryptoSlate · · Single source

Analysis and references

Impact analysis

The overhaul affects ENS token holders and the secondary market for .eth names, since protocol fee rules and domain ownership sit at the centre of the upgrade. Uncertainty around a one-way migration can influence how domain marketplaces and traders transfer .eth names during the transition period.

References

Tags

#ENS#.eth#ENSv2#Ethereum Name Service#protocol upgrade#domain names

#11
Crypto
7.5

Interactive Brokers Launches Ethereum Staking Through Zerohash

Background

Background

zerohash is a crypto, stablecoin and tokenization infrastructure provider that supplies banks, brokerages and fintechs with the technology to launch digital asset products, including regulated crypto trading through a single API integration. Ethereum staking involves depositing ETH to activate a validator, a participant in Ethereum's consensus protocol that proposes and verifies blocks.

Interactive Brokers (NASDAQ: IBKR) has launched cryptocurrency staking, with infrastructure provided by digital asset firm zerohash. Initially, eligible Interactive Brokers LLC clients can stake ETH through the Staking Dashboard in IBKR's Client Portal, where they submit staking and unstaking requests and monitor pending activity.

IBKR CEO Milan Galik said the staking service lets clients earn additional yield on the crypto assets they hold, similar to the firm's stock yield enhancement program. It extends ETH staking access to a mainstream brokerage customer base.

Rewards, when earned, are credited as additional staked cryptocurrency and can compound while the position remains staked. IBKR currently supports trading in 20 digital assets and accepts deposits in USDC, RLUSD and PYUSD.

google_news · The Defiant · · 5 sources

Analysis and references

Impact analysis

The main transmission channel is distribution: a large traditional brokerage now funnels eligible client ETH into staking through a third-party infrastructure provider, which could add to the pool of ETH committed to staking and expand demand for staking and custody infrastructure among regulated intermediaries. The service is initially limited to eligible Interactive Brokers LLC clients, so any market effect depends on how widely it is adopted.

References

Tags

#ethereum-staking#interactive-brokers#zerohash#tradfi#institutional-adoption

#12
Crypto
7.5

Binance Sets Stricter Rules for Market Makers and Token Issuers

Background

Background

Market makers are firms that supply liquidity to crypto markets by continuously buying and selling assets so order books have enough depth for other participants to trade against. Token issuers are the projects and foundations behind crypto assets, and exchanges such as Binance set their own requirements for listing and for the market-making arrangements that support a newly listed token.

Binance introduced stricter rules governing market makers and token issuers, tightening the standards these parties must meet to operate and list on the platform.

Binance is the largest crypto exchange by trading volume, so changes to how it screens the market makers that supply order-book liquidity and the issuers that seek listings directly affect how tokens reach one of the market's main trading venues.

google_news · CoinMarketCap · · Single source

Analysis and references

Impact analysis

The rules target the order-book liquidity that Binance market makers provide and the listing economics for token issuers, so the transmission runs through liquidity depth and listing terms on Binance rather than through a specific asset; projects and trading firms that cannot meet the tighter standards may face changes to their arrangements on the venue.

References

Tags

#binance#market-makers#token-listings#exchange-policy#market-structure

#13
7.5

Bybit CEO: Bybit stock trading will later support collateral, lending and broker position transfers

Background

Background

Bybit's Unified Trading Account is the venue's all-in-one account mode, giving traders access to spot, spot margin, perpetuals, futures and options without switching accounts. Tokenized equities — 1:1 backed digital representations of US stocks and ETFs tradable around the clock across chains — have become a related but distinct route for bringing equity exposure onto crypto venues.

Bybit CEO Ben Zhou announced the launch of Bybit Direct Stock during the "Make Your Move" global keynote on October 13, giving users direct access to more than 1,000 US market trading pairs and assets, with real stock ownership. He said the stock assets will be held in Bybit's Unified Trading Account (UTA), and that the platform plans to later support equities as trading collateral, introduce stock lending, and allow users to transfer in positions held at brokers such as Futu and Moomoo.

The launch puts real US equity ownership inside a crypto exchange's unified account, and the stated roadmap toward equity collateral, stock lending and broker transfers-in would place a crypto-native platform in direct competition with traditional brokerages. The collateral, lending and transfer features are forward-looking plans announced by the CEO, not live functionality.

According to Zhou, the stock assets are custodied within the Unified Trading Account, which Bybit describes as centralizing spot, futures and options trading in one account. The planned broker transfer-in covers positions held at Futu and Moomoo, both of which support transferring US stocks out to another brokerage via methods such as DTC free-of-payment (FOP) transfers.

telegram · theblockbeats · · 2 sources

Analysis and references

Impact analysis

If equities are eventually accepted as collateral inside Bybit's unified account, margin and liquidation dynamics could transmit between equity and crypto positions, linking crypto order books more directly to US equity market structure and to established brokerage transfer and custody rails such as DTC FOP. The immediate exposure is concentrated in Bybit's own UTA users, since the collateral, lending and transfer-in functions remain announced plans.

References

Tags

#Bybit#tokenized-equities#exchange-product#market-structure#collateral

#14
7.5

SkyAI(@skyaico), which held 2M $SOL ($214M), has been dumping $SOL heavily recen

Background

Background

A cash tender offer is a public offer to buy shares directly from holders for cash at a fixed price and for a limited time. A pre-funded warrant grants the holder the right to acquire shares at a nominal exercise price, with the funding effectively provided at issuance; such instruments are commonly issued alongside financing transactions. SkyAI's disclosure follows a period in which several listed companies accumulated large SOL positions as corporate treasury assets.

SkyAI, a company that held about 2 million SOL (~$214M), has disclosed that it sold 1 million SOL (~$107M) and has said it may continue selling. On-chain data shows the entity transferred out 1.57 million SOL (~$175M) during October, against an average purchase price of roughly $194 per SOL, implying losses of more than $160M.

Management's statement that it may keep selling matters because SkyAI was among the largest corporate holders of SOL, and the disclosed sales are an explicit, repeated source of token supply. The company is simultaneously running a $3.00-per-share cash tender offer, tying the unwind to its own equity.

The Arkham entity page tracking the addresses is labeled "Sharps Technology," while the company is referred to as SkyAI. Transferred-out tokens are on-chain movements and are not by themselves proof of sale; the confirmed disposal figure disclosed by the company is 1 million SOL. The tender offer at $3.00 per share covers common stock and pre-funded warrants.

telegram · lookonchainchannel · · Single source

Analysis and references

Impact analysis

The disclosed sales and the company's stated willingness to keep selling put identifiable sell-side supply into SOL spot markets, where this holder's treasury position was one of the largest corporate ones. The separate $3.00-per-share cash tender offer affects the equity and pre-funded warrant side rather than the token, and the size of the unrealized loss is likely to weigh on sentiment toward corporate SOL treasury strategies.

References

Tags

#solana#corporate-treasury#large-holder-selling#on-chain-data#market-structure