Grayscale’s Zcash ETF (ZCSH) plans a 3-for-1 forward share split, according to a filing with the US Securities and Exchange Commission.

At the close of trading on Sept. 28, shareholders will receive two extra shares for each one they hold, according to the filing.

The forward split is expected to decrease the price per share of the fund, according to a Grayscale press release, with a proportionate increase in the number of shares outstanding.

Hypothetically, this means that if you owned 10 shares valued at $300 each for a total $3,000 before the split, afterward you will own 30 shares valued at $100 each for an unchanged total of $3,000, the release said.

The split will make the ETF more accessible to investors, as the token has increased in value by about 2,800% over the last year and the price per unit was considered too high.

Related: Zcash targets November for NU7 mainnet upgrade with 25-second blocks

Cointelegraph reported on Thursday that Zcash (ZEC), a cryptocurrency that enables users to make shielded transactions that conceal addresses and transaction amounts using zero-knowledge proofs, had gained about 20% over 24 hours as Paradigm co-founder Matt Huang disclosed that the crypto investment firm made an unspecified purchase of ZEC.

Huang described Zcash as a “private complement to Bitcoin” and also backed its developer fund, arguing that long-term funding remains important as AI-driven cyber capabilities and quantum computing advance.

Zcash’s ZEC token climbed as high as $1,521 early Friday, The Block reported, in what would be considered a new effective all-time high, before falling back slightly.

Magazine: Who is legally liable when an AI agent goes rogue?

Cointelegraph is committed to independent, transparent journalism. This news article is produced in accordance with Cointelegraph’s Editorial Policy and aims to provide accurate and timely information. Readers are encouraged to verify information independently.

Anthropic has selected Accenture as its first “embedded evaluator,” a move tied to CEO Dario Amodei’s recent proposal to slow frontier AI development while safeguards catch up. The partnership is designed to put independent testing closer to how advanced AI systems are built and deployed—an approach Anthropic says it believes is necessary to reduce the risk of catastrophic harm.

The announcement follows Amodei’s call for a slower, safer pace of AI progress published Sept. 12. His argument centers on the way AI systems can improve other AI systems: as models become better at creating next-generation models, development can accelerate through recursive self-improvement—potentially outpacing human oversight.

Key takeaways

• Anthropic is partnering with Accenture to act as its first embedded evaluator, conducting model evaluations and red-teaming.
• Amodei’s Sept. 12 proposal emphasized independent evaluators with “employee-like” access to test safeguards more effectively.
• Anthropic expects to invest directly in this early phase because a dedicated funding mechanism for independent evaluation does not yet exist.
• The deal is non-exclusive, and Anthropic plans to name additional evaluators in the coming weeks.
• Both Anthropic and Accenture expect to invest at least $1 billion over the next five years, according to Anthropic’s announcement.

Why “embedded evaluation” is at the center of Anthropic’s safety plan

Amodei’s proposal framed a core problem: AI progress has been advancing “drastically faster” as systems improve their ability to build the next generation of AI. In his account, that feedback loop—often described as recursive self-improvement—could allow powerful systems to evolve faster than society’s ability to understand and control them.

To address that mismatch, the proposal’s first step calls for independent evaluators granted access comparable to internal teams. Anthropic says it had already committed to this direction, and the Accenture partnership is positioned as an early, concrete way to operationalize it.

In its announcement, Anthropic described embedded evaluation as work that includes evaluating and red-teaming models, running alignment assessments, and testing safeguards. Because embedded evaluation is new, Anthropic also cautioned that the specific implementation details are still being worked out.

Accenture will lead the first embedded evaluation effort

Anthropic said it is moving toward its earlier commitment by partnering with Accenture’s AI business unit, Faculty. The company’s stated goals for the embedded evaluator include:

• Evaluating models and performing red-team testing
• Conducting alignment assessments
• Testing model safeguards

Anthropic and Accenture each expect to invest at least $1 billion in the project over the next five years, according to the announcement. That investment commitment underscores that embedded evaluation is being treated not as a short-term audit, but as an ongoing capability that must scale alongside frontier model development.

Accenture, for its part, described Faculty as experienced in testing and evaluating models for major AI labs and in building complex systems designed to be safe and ethical by design. Earlier reporting also highlighted the partnership effort and Accenture’s focus on building evaluation capacity, as referenced by Yahoo Finance.

Funding the evaluators—and keeping independence—remains a challenge

One of the more operational elements of Anthropic’s announcement is how it expects the work to be financed. Anthropic noted there is currently no established system for funding independent evaluation. It suggested that in the longer term, pooled or government sources could support this kind of work, but it also acknowledged that time is critical.

Given the urgency, Anthropic said it will fund Accenture’s embedded evaluation directly in this phase. The company also emphasized that the partnership is non-exclusive, adding that it expects to announce additional evaluators in the coming weeks. For investors and builders, this matters because it suggests Anthropic is trying to avoid a single-evaluator dependency and instead build a broader ecosystem of independent oversight.

Whether embedded evaluation can remain truly independent while embedded in production processes is likely to become a key question for the industry to watch. Anthropic’s plan to appoint multiple evaluators may help mitigate that risk by introducing more than one perspective and reducing the chance that evaluation becomes a narrowly defined internal function.

Industry reaction highlights the policy tension around slowing AI

Amodei’s Sept. 12 proposal did not arrive in a vacuum. Reports from the period show a split among major tech leaders on whether slowing AI development is the right response to safety concerns. OpenAI CEO Sam Altman and SpaceX CEO Elon Musk were reported as responding positively to Amodei’s framing, while Nvidia CEO Jensen Huang argued that regulation aimed at slowing development was not necessary.

That tension reflects a broader debate that extends beyond one company: some leaders view additional constraints as the only way to ensure safeguards can keep up, while others believe the focus should remain on technical safety measures rather than pacing controls. Anthropic’s next step—securing an embedded evaluator and funding it heavily—signals that, at least for this phase, the company is pursuing a concrete evaluation mechanism rather than relying solely on public policy or abstract promises.

What to watch next

As Anthropic works to finalize how embedded evaluation will operate in practice, the immediate question for the sector is whether the approach can scale beyond a single pilot evaluator and remain credible as a form of independent oversight. Readers should also watch the timeline and criteria for Anthropic’s additional evaluators, since the non-exclusive plan may determine whether embedded evaluation becomes an industry standard or remains a bespoke effort.

Onchain real-world assets have reached $34.18 billion as of September 15, 2026, after growing 85.2% since the start of the year, while only around 12% of tracked tokenized capital is being used in onchain financial applications.

Binance Research published the figures on September 18 in its“The RWA Activation Era” report, using DefiLlama data and its own methodology to compare asset issuance with onchain use. Bonds and money market funds remained the largest category at $18.29 billion, while tokenized equities reached $4.43 billion after rising 390.4% year to date.

The report separates tokenization into two measures. Its Programmable Asset Ratio, or PAR, compares tokenized value with the size of the underlying market. Capital Activation Rate, or CAR, measures how much eligible tokenized value is deployed in liquidity pools, lending markets, collateral systems and other verified onchain applications.

RWA growth is being led by bonds and tokenized stocks

Bond and money market funds generated 54.7% of this year’s increase in tracked RWA assets, Binance Research said. Equities contributed another 22.4%, meaning the two categories produced more than three-quarters of the added market value through September 15.

Other categories expanded at slower rates. Gold and commodities rose 46.6% year to date, private credit increased 43.6%, and real estate gained 17.9%, according to the report.

Tokenized equities posted the fastest percentage growth among the main categories tracked by Binance Research. Their share of RWA assets rose from 4.9% to 13.0%, even though the $4.43 billion onchain balance represented only 0.0029% of the $151.9 trillion listed-equity reference market used in the report.

Advertisement

Across all covered markets, Binance Research estimated that only around 0.01% of the underlying asset base has been tokenized. Bond and money market funds had an indicative PAR of 0.0171%, leaving their onchain share small compared with the traditional markets they represent.

The latest figures extend an expansion documented earlier in the year. In June, Binance Research had recorded rapid growth in tokenized stocks while bond and money market funds added billions of dollars in onchain value.

Most tokenized capital remains outside DeFi applications

The report found an overall CAR of roughly 12%, meaning close to $12 of every $100 in qualifying tokenized asset value was deployed in tracked financial applications. Binance Research used the measure to separate issued assets from assets being used as liquidity, collateral or lending capital.

A separate DeFiLlama-based review published earlier in September produced a similar result.As crypto.news reported, around $3.79 billion of a $34.6 billion tokenized RWA market was deployed in protocols at the time, leaving roughly 89% outside the applications covered by the dataset.

Advertisement

Utilization varies sharply by asset type. Binance Research put private credit CAR at 49.67%, the highest among its tracked categories. Equity CAR rose from 1.95% at the start of the year to 7.54% by September 15.

Within tokenized-equity DeFi activity, liquidity pools accounted for 65.4% of deployed value and lending represented 28.1%. Together, the two uses made up 93.5% of equity DeFi total value locked measured by the report.

Product-level data can look very different from the market average. The crypto.news review of DefiLlama data found BlackRock’s BUIDL at 0.64% utilization, Franklin Templeton’s BENJI at 0%, and Circle’s USYC at 0.52%. Centrifuge’s JAAA and Re Protocol’s reUSD both had utilization above 97% in the same dataset.

U.S. rules now give tokenized equities a limited onchain route

One day before Binance Research published its report, the U.S. Securities and Exchange Commission approved a temporary framework for limited onchain trading of tokenized National Market System stocks.

Advertisement

The SEC’s September 17 Innovation Exemption gives qualifying Tokenized Securities Venues conditional relief from the Exchange Act definition of an exchange. Related relief covers certain liquidity providers supplying proprietary capital through permissioned automated market makers and liquidity pools.

The five-year exemption comes with limits. Tokenized NMS stocks must provide the same rights and privileges as corresponding traditional shares, including voting and dividend rights where applicable. Issuers must be able to object when an unaffiliated third party wants to make a tokenized version of their stock available on a qualifying venue.

Trading venues must use auditable public smart contracts on public permissionless distributed ledgers, follow trading halts in the underlying security, maintain records and publish required transaction information. Anti-fraud and anti-manipulation provisions continue to apply.

The regulator is seeking public comment while it considers longer-term rules. SEC Chairman Paul Atkins said the exemption would permit trading in a permissioned environment “while the Commission considers the need for additional action” on onchain stock trading.

Advertisement

Market infrastructure is moving tokenized assets toward use

Institutional projects are putting tokenized securities into trading, collateral and settlement workflows. On September 16, DTCC said Ondo Finance subsidiary Oasis Pro Markets had joined Fund/SERV, becoming the platform’s first tokenization member. DTCC said Fund/SERV processes more than 85% of U.S. mutual fund transaction activity.

DTCC had already completed production transactions using DTC-tokenized assets on July 15. Participating firms used tokenized securities in Treasury repo, equity delivery-versus-payment, securities lending, collateral pledge and central-counterparty margin workflows, according to the company.

The DTC tokenization service remains scheduled for an October 2026 launch. DTCC said tokenized versions of DTC-custodied securities are designed to retain the same ownership rights, entitlements and investor protections as their traditional forms.

DeFi lenders have built separate channels for RWA collateral. Aave launched Horizon in August 2025 for qualified borrowers seeking stablecoin liquidity against tokenized assets. By February 2026, Aave Labs said deposits had exceeded $440 million. 

Advertisement

Aave plans a dedicated RWA credit hub on Avalanche where eligible institutions could borrow USA₮ against approved tokenized financial assets. Aave’s governance materials had previously specified that a dedicated RWA hub would follow the initial Avalanche V4 deployment through a separate governance process.

Binance Research sees issuance and utilization moving separately

For tokenized equities, Binance Research used three 2030 scenarios from an earlier report: approximately $61 billion, $349 billion and $987 billion in tokenized equity value. The firm presented the figures as scenario ranges, not company-confirmed future market values.

Under its $349 billion base scenario, Binance Research estimated a PAR of 0.23%. Its sensitivity analysis showed that moving equity CAR from 10% to 20% at that asset level would increase deployed capital from $34.94 billion to $69.87 billion without requiring more tokenized supply.

The report calls the next stage an “RWA Activation Era,” referring to tokenized assets becoming usable in exchanges, lending and collateral markets. Current figures still show a large difference between asset value represented onchain and capital deployed in those applications.

Advertisement

Binance Research cited earlier platform data showing 58.5% of early bStocks users used perpetuals or direct equities as well. The report said future adoption will depend on whether distribution channels and financial applications convert access to tokenized assets into recurring liquidity and financing activity.

 

Anthropic said it has chosen Accenture as its first embedded evaluator to help slow the pace of AI development as proposed last week by CEO Dario Amodei.

Amodei published a three-step proposal on Sept. 12 to slow the development of AI and allow safeguards to be put in place amid warnings of a potential for catastrophic harm in the wake of swift, unchecked development.

OpenAI CEO Sam Altman and SpaceX CEO Elon Musk responded positively to Amodei’s proposal, although Nvidia CEO Jensen Huang did not, arguing that such regulation was not necessary.

Still, Amodei wrote in his proposal that “AI has been advancing drastically faster, driven primarily by AI’s growing ability to build the next generation of AI. This dynamic is called recursive self-improvement,” and “left unchecked, it could outrun our ability to understand and control these systems.”

Related: Anthropic chief urges slowdown in AI development to safer pace

The first step in Amodei’s proposal is having independent evaluators with employee-like access, to which Anthropic had already unilaterally committed, Amodei wrote.

Anthropic announced a move toward fulfilling this commitment on Friday by partnering with Accenture and its AI business Faculty in “evaluating and red-teaming models, conducting alignment assessments and testing model safeguards.”

Details of how this will happen are still being worked out, as embedded evaluation is new, the company said.

Anthropic and Accenture each expect to invest at least $1 billion in the project over the next five years, according to the announcement.

Anthropic will fund work directly amid urgency

Anthropic noted that there is also no existent system for funding independent evaluation, so long-term funding should come from pooled or government sources, though given the urgency of the work, Anthropic will fund Accenture’s work directly.

The partnership is non-exclusive, and Anthropic expects to announce other evaluators in forthcoming weeks with which it will also work.

Accenture’s Faculty is expert in testing and evaluating models for some of the world’s leading AI labs and building complex AI systems that are safe and ethical by design, the company said on Friday.

“Embedded evaluation is an emerging area, and we look forward to partnering with Anthropic to help accelerate the development of embedded evaluators, which we see as an important part of the safety landscape going forward,” said Julie Sweet, chair and CEO of Accenture. 

Magazine: Why are AI’s biggest companies suddenly asking to slow down?

Polymarket has come under renewed scrutiny after a Sept. 19 report said fraudsters used stolen debit cards on its U.S. platform in February to attempt at least $10 million in illicit withdrawals and wagers.

The Wall Street Journal reported that criminals connected stolen cards to thousands of Polymarket US accounts, funded them and then tried to move the money through trading before withdrawing it to cards or accounts they controlled.

Fraudsters flooded Polymarket with stolen debit cards. The CEO’s response shocked some employees, showing how the betting app valued growth at all costs. https://t.co/1bIbKrO0uW

— The Wall Street Journal (@WSJ) September 20, 2026

Advertisement

Polymarket reportedly saw fraud rejection rates top 80%

At the peak of the February attack, payment processor Checkout.com rejected more than 80% of deposits it handled for Polymarket as fraudulent, according to the Journal. The newspaper compared that rate with an industry level of roughly 1%.

The 80% figure has not been independently confirmed by Checkout.com in a public statement reviewed for this report. Checkout.com does provide merchants with fraud-scoring, transaction-filtering and authentication tools, while its current service terms state that merchants remain responsible for deciding whether transactions are accepted or canceled.

Current and former employees told the Journal that compliance workers escalated concerns about the surge to Polymarket CEO Shayne Coplan. According to people cited by the newspaper, Coplan responded: “Just keep growing and pay a fine if regulators ever find out.”

Polymarket has not publicly confirmed that Coplan made the remark. The company told the Journal that it maintains procedures to identify and respond to suspicious activity and remains committed to cooperating with regulators and law enforcement.

The $10 million figure represents the amount fraudsters allegedly tried to move, not a confirmed loss suffered by customers or Polymarket. Public reporting reviewed for this article does not provide a final amount successfully withdrawn through the February scheme.

Card restrictions helped bring fraud rates down by May

The Journal reported that elevated fraud continued for several months after February, though rejection rates did not return to the peak recorded during the first wave. By May, fraud rates had reportedly moved back toward normal industry levels after Polymarket limited how many debit cards users could connect to their accounts and brought in Riskified as an outside antifraud provider.

An earlier report from The Information had separately described prediction-market operators, including Polymarket, strengthening card-fraud controls after criminals used stolen payment credentials and fake identities to create accounts. Visa reportedly pushed payment processors to tighten screening as disputed transactions increased.

Riskified provides automated fraud-decision systems used to identify suspicious card activity before merchants approve transactions. Public material from the company describes its service as combining machine-learning risk scoring with merchant transaction controls, but Riskified has not publicly disclosed Polymarket-specific fraud numbers.

The Journal further reported that Polymarket initially required some withdrawals to return to the same payment source that had funded an account. The platform later loosened that restriction, according to the report, which cited employees who raised concerns about financial-crime risks.

Polymarket’s current U.S. rulebook gives the exchange authority to restrict accounts, place customers into liquidation-only status and take other steps to protect customers and market integrity. The CFTC filing containing the March 20 version was certified in April.

Polymarket US operates under CFTC-regulated QCX

Polymarket US is legally separate from the company’s international blockchain-based prediction market.

The Commodity Futures Trading Commission’s current register lists QCX LLC, doing business as Polymarket US, as a designated contract market. QCX received its designation in July 2025 before operating under the Polymarket US name.

As previously explained in Polymarket’s two-platform structure, U.S. customers trade through the federally regulated exchange, while the international product uses separate blockchain infrastructure and access rules.

The regulatory status differs from Polymarket’s position in 2022, when the CFTC ordered the company to pay a $1.4 million civil penalty for offering event-based binary options without operating through a registered market. The settlement required Polymarket to wind down noncompliant markets and cease the violations cited in the order.

The Journal reported that the CFTC is now investigating issues connected with Polymarket and that employees were instructed to preserve documents involving the February fraud incident and other matters.

No new public CFTC enforcement release specifically addressing the February stolen-card episode was located as of Sept. 20. The reported investigation should therefore be treated as an ongoing inquiry described by the Journal, not a finding that Polymarket violated federal law.

Separate congressional scrutiny was already underway. On May 22, the House Committee on Oversight and Government Reform requested records from Polymarket concerning identity verification, suspicious activity, geographic restrictions and referrals to U.S. authorities.

The committee specifically requested documents showing the number and disposition of suspicious-activity referrals since Jan. 1, 2024. That inquiry centered on insider trading and sensitive information, not the stolen-card scheme reported this weekend.

Polymarket has expanded investigations and finance teams

Since the February incident, Polymarket has built out its internal investigation and management functions. Shana Bautista, a former FBI investigator, joined as global head of investigations and intelligence. Reuters reported in August that the company uses blockchain analytics, machine learning and trading surveillance to identify anomalous behavior.

Polymarket’s own market-integrity page says the company has referred more than 90 accounts to law enforcement and supplied authorities with details involving more than 315 wallets. The figures are company-reported and do not relate exclusively to payment-card fraud.

Federal authorities have publicly acknowledged cooperation in at least one separate case. In April, the U.S. Attorney’s Office for the Southern District of New York said Polymarket cooperated with investigators in the case of an Army service member accused of using classified information to trade event contracts.

The CFTC filed a parallel insider-trading complaint alleging that the defendant earned more than $404,000 trading a market related to the capture of Nicolás Maduro. The case remains separate from the February payment-card allegations.

Polymarket added another senior executive on Sept. 10 when it named Warren Jenson its first chief financial officer. The CFO appointment, Jenson previously served in senior finance roles at Amazon, Electronic Arts, Delta Air Lines and Nielsen.

The company said Jenson will oversee finance, capital strategy and long-range planning. Polymarket did not announce an IPO timetable when it appointed him, although the Journal reported that the company is preparing itself for a potential public listing.

Polymarket has been raising large amounts of private capital in parallel. ICE, the parent of the New York Stock Exchange, disclosed a further $600 million cash investment in March after investing $1 billion in 2025.

As recent Polymarket funding coverage reported, the company has separately been seeking roughly $1 billion in new capital at a valuation near $21 billion. The financing has not been presented by Polymarket as a formal IPO filing.

Separate security incidents added pressure during 2026

Payment fraud has not been the company’s only security issue this year. In June, Polymarket confirmed that a compromised third-party vendor injected malicious code into its frontend for some users. The company said it removed the affected dependency, contained the incident and would reimburse affected customers.

Blockchain investigators later estimated losses at roughly $3.1 million across 11 wallets. AMLBot said stolen assets were moved from Polygon to Ethereum after the malicious activity.

The June event was technically separate from the February stolen-card activity. It involved compromised web infrastructure and wallet interactions, while the earlier scheme reportedly relied on stolen debit-card credentials and account funding.

The Journal reported another account-security episode in July involving nearly 500 users. According to its account, attackers used stolen personal information to access existing accounts and linked payment methods through an engineering weakness. Polymarket reportedly agreed to cover affected losses.

No public Polymarket notice reviewed for this report provides an independently confirmed loss total from that July incident.

The company’s current position is that its fraud controls have been strengthened and that it works with law enforcement on suspicious activity. The Journal’s report says February’s payment-fraud rate had returned to industry norms by May after debit-card restrictions and the Riskified deployment.