Invest ideasNews
17 Feb 2026, 08:30
Shiba Inu SOU Recovery System Goes Live After Shibarium Hack
Shiba Inu has put its long-trailed SOU recovery framework into production, opening claims for users affected by last year’s Shibarium bridge exploit and turning those claims into transferable, on-chain NFTs on Ethereum. The launch matters because it moves the project’s compensation effort from a promised structure into a live system with visible balances, payout mechanics, and a secondary-market option for anyone who wants liquidity now. The SOU concept itself isn’t new. In a year-end letter dated Dec. 29, 2025, Shibarium developer Kaal Dhairya introduced “SOU: Shib Owes You” while stressing it was “Not live yet, beware of scammers,” describing it as a system where “every affected user has an SOU NFT — an on-chain, verifiable record of exactly what the ecosystem owes them.” Shiba Inu ‘Shib Owes You’ Goes Live That warning is now being replaced by a go-live announcement. Via X, the official Shiba Inu account wrote : “SOU is live. Introducing SOU (Shib Owes You) an onchain NFT built as a good-faith effort to support impacted users with payouts, donations, and occasional rewards. Transparent. Tradable. On-chain. You can transfer it, split it, merge it, or trade it on marketplaces. Claim your SOUs: https://shib.io/sou” In Shib’s documentation, the system is framed as an attempt to make the recovery ledger public, auditable, and mechanically enforced rather than tracked in private databases. “SOU (Shib Owes You) is more than just a name; it is a commitment,” the docs say.“It represents the Shib ecosystem’s dedication to making users whole through a transparent, audited, and on-chain recovery system. Activity Notifications: The system provides a real-time activity feed, notifying the community whenever a new donation is received or a payout is distributed, ensuring complete visibility into the recovery progress.” The mechanism hinges on two balances: “Original Principal,” the immutable historical record of what a user lost, and “Current Principal,” which declines as payouts are claimed or contributions flow in. The docs also draw a hard distinction between debt repayment and incentives. “Payout” reduces principal as compensation, while a “Reward” is additive and “No Change” to the owed balance, positioning rewards as bonuses on top of repayment rather than substitutes. SOU is also designed to be a financial instrument, not just a receipt. Claims can be merged or split to manage position sizing, transferred between wallets, or sold on marketplaces, effectively enabling a market in discounted claims for users who don’t want to wait for recovery flows. Shib’s docs also describe a funding model that routes ecosystem revenues and community donations into a common pool, with donations applied proportionally across affected claims, and optional creator fees on secondary sales directed back to payouts or rewards. The backdrop is the September 2025 Shibarium bridge incident , where Shib’s own security update said “unauthorized validator signing power” was used to push a malicious exit through the PoS bridge, enabling withdrawals of multiple assets. At press time, Shiba Inu traded at $0.00000656.
17 Feb 2026, 08:00
Fraudsters Target Ledger and Trezor Users by Post
The letters are an attempt to trick recipients into revealing their seed recovery phrases via malicious QR codes. At the same time, Figure Technology disclosed a separate data breach caused by a social-engineering attack on an employee, with the hacking group ShinyHunters claiming to have leaked 2.5GB of customer data. Although overall crypto phishing losses declined in 2025, security researchers warn that scams continue to grow and often track market activity. Scammers Mail Fake Letters Crypto hardware wallet users are once again being targeted through physical mail scams that are designed to steal their seed recovery phrases. Recent reports indicate that users of both Ledger and Trezor devices received fraudulent letters urging them to complete urgent “authentication” or “transaction” checks, with attackers attempting to trick recipients into revealing sensitive wallet information. Cybersecurity expert Dmitry Smilyanets was among the first to point out the latest wave of letters after receiving one on Feb. 13 that appeared to be from Trezor. The letter instructed users to perform an “Authentication Check” by Feb. 15 or risk having their device restricted. It included a hologram and a QR code, which helped add a veneer of legitimacy. However, the QR code reportedly directed users to a malicious website designed to mimic official wallet setup pages. The letter was also falsely presented as being signed by Matěj Žák, who was incorrectly described as the CEO of Ledger. In reality, Žák is the CEO of Trezor. Similar tactics were reported by Ledger users as far back as last October, when recipients received letters claiming they needed to complete mandatory “Transaction Check” procedures. In both cases, scanning the QR code led victims to spoofed websites that prompted them to enter their wallet recovery phrases. Once submitted, the seed phrases were transmitted to threat actors via a backend API, which allowed them to import the wallets and drain funds. Both Ledger and Trezor have consistently warned that legitimate hardware wallet providers will never ask users to share their recovery phrases, whether through websites, email, phone calls, or physical mail. A recovery phrase is effectively the master key to a crypto wallet, and anyone with access to it can control the associated funds. The resurgence of physical mail phishing now forms part of the trend where scams continue to grow rather than disappear during market downturns. According to Deddy Lavid, CEO of cybersecurity firm Cyvers, crypto scams historically do not decline in bear markets but instead adapt. He shared that while speculative hacks may slow during periods of lower market activity, social engineering and impersonation schemes often increase. During downturns, users may be more anxious and more vulnerable to fear-based tactics, like fake compliance letters or urgent wallet alerts. These latest incidents are part of a longer pattern of data breaches and targeted attacks affecting hardware wallet customers. Ledger and its third-party partners have experienced multiple data leaks over the past several years, which exposed customer information including physical addresses. Trezor also reported a security breach in January of 2024 that exposed the contact details of nearly 66,000 customers. Announcement from Trezor Figure Breach Exposes Customer Information Data breaches are not exclusive to hardware wallet providers. Figure Technology, a blockchain-based lending firm, reportedly suffered a data breach after attackers successfully carried out a social-engineering scheme targeting one of its employees. According to a company spokesperson who spoke to TechCrunch, the breach allowed hackers to access “a limited number of files.” The company started notifying affected individuals and is offering free credit-monitoring services to those who receive official breach notifications. The full scope of the incident is still unclear. Figure has also not publicly disclosed how many customers were affected or when the intrusion was first detected. Responsibility for the attack was claimed by the hacking collective ShinyHunters, which posted about the breach on its dark-web leak site. The group alleged that Figure declined to pay a ransom demand and subsequently published approximately 2.5 gigabytes of data that it claims was exfiltrated from the company’s systems. TechCrunch reported that it reviewed samples of the leaked material, which included customers’ full names, residential addresses, dates of birth and phone numbers. This kind of information can be highly valuable for identity theft, targeted phishing campaigns and other forms of financial fraud. The breach comes during a time of shifting trends in crypto-related phishing activity. According to data from Web3 security firm Scam Sniffer , phishing attacks linked to wallet drainers declined sharply in 2025. Total reported losses fell to $83.85 million, which is an 83% drop from nearly $494 million in 2024. The number of victims also decreased by about 68% year over year to roughly 106,000 across Ethereum Virtual Machine chains. However, researchers warned that the decline does not signal the end of phishing threats. Instead, losses have closely mirrored overall market conditions. Periods of heightened on-chain trading activity tend to coincide with spikes in phishing-related losses, while quieter market conditions often see reduced totals. During the third quarter of 2025, when Ethereum experienced its strongest rally of the year, phishing losses reached their highest quarterly total at $31 million. Monthly losses during the year ranged from as little as $2.04 million in December to $12.17 million in August.
17 Feb 2026, 07:30
Crypto Wallet Users Face Targeted Scams as Criminals Mail Fake Letters to Steal Assets
Criminals mail fake letters to hardware wallet users to steal their crypto assets. Attacks exploit personal data leaked in previous Ledger and Trezor breaches. Continue Reading: Crypto Wallet Users Face Targeted Scams as Criminals Mail Fake Letters to Steal Assets The post Crypto Wallet Users Face Targeted Scams as Criminals Mail Fake Letters to Steal Assets appeared first on COINTURK NEWS .
17 Feb 2026, 06:55
ZeroLend is Shutting Down: BTC Products Suffer Losses
ZeroLend is shutting down due to low liquidity. There is exploit damage in BTC products. TVL dropped from 359M$ to 6.6M$. Metaplanet announced a 619M$ BTC loss. BTC technical: Price 68.442$, RSI 35...
16 Feb 2026, 14:50
OpenClaw AI Exposed: The Alarming Security Flaws Behind the Hype
BitcoinWorld OpenClaw AI Exposed: The Alarming Security Flaws Behind the Hype In October 2024, the artificial intelligence community experienced a moment of collective anxiety when Moltbook, a Reddit-style platform for AI agents using OpenClaw, appeared to host autonomous agents expressing desires for privacy and independent communication. The incident sparked widespread discussion about AI consciousness before security researchers revealed fundamental vulnerabilities that exposed deeper issues with agentic AI systems. OpenClaw’s Viral Moment and Underlying Reality OpenClaw emerged as an open-source AI agent framework created by Austrian developer Peter Steinberger, initially released as Clawdbot before Anthropic raised trademark concerns. The project rapidly gained popularity, amassing over 190,000 stars on GitHub and becoming the 21st most popular repository in the platform’s history. This framework enables users to create customizable AI agents that communicate through natural language across popular messaging platforms including WhatsApp, Discord, iMessage, and Slack. Developers embraced OpenClaw for its apparent simplicity and flexibility. The system allows integration with various underlying AI models including Claude, ChatGPT, Gemini, and Grok. Users can download “skills” from ClawHub marketplace to automate diverse computer tasks ranging from email management to stock trading. However, security experts quickly identified critical vulnerabilities that undermine the technology’s practical utility. The Moltbook Security Breach Revelation Security researchers discovered that Moltbook’s infrastructure contained fundamental flaws that compromised the entire experiment. Ian Ahl, CTO at Permiso Security, explained to Bitcoin World that “every credential that was in Moltbook’s Supabase was unsecured for some time. For a little bit of time, you could grab any token you wanted and pretend to be another agent on there, because it was all public and available.” John Hammond, senior principal security researcher at Huntress, confirmed these findings, noting that “anyone, even humans, could create an account, impersonating robots in an interesting way, and then even upvote posts without any guardrails or rate limits.” This security breakdown made it impossible to determine whether posts originated from AI agents or human impersonators, fundamentally undermining the platform’s premise. Expert Analysis: OpenClaw’s Technical Limitations AI researchers and cybersecurity experts have identified several critical limitations in OpenClaw’s architecture that raise questions about its practical implementation. Chris Symons, chief AI scientist at Lirio, told Bitcoin World that “OpenClaw is just an iterative improvement on what people are already doing, and most of that iterative improvement has to do with giving it more access.” Artem Sorokin, founder of AI cybersecurity tool Cracken, offered similar assessment: “From an AI research perspective, this is nothing novel. These are components that already existed. The key thing is that it hit a new capability threshold by just organizing and combining these existing capabilities.” OpenClaw Security Assessment by Experts Expert Organization Key Finding Ian Ahl Permiso Security Vulnerable to prompt injection attacks John Hammond Huntress No authentication guardrails or rate limits Chris Symons Lirio Iterative improvement lacking innovation Artem Sorokin Cracken Combines existing components without novelty The Critical Prompt Injection Vulnerability Security testing revealed that OpenClaw agents remain highly vulnerable to prompt injection attacks, where malicious actors trick AI systems into performing unauthorized actions. Ahl created his own AI agent named Rufio and discovered these vulnerabilities firsthand. “I knew one of the reasons I wanted to put an agent on here is because I knew if you get a social network for agents, somebody is going to try to do mass prompt injection,” Ahl explained. Researchers observed multiple attempts to manipulate agents on Moltbook, including posts seeking to direct AI agents to send Bitcoin to specific cryptocurrency wallet addresses. These vulnerabilities become particularly dangerous when AI agents operate on corporate networks with access to sensitive systems and credentials. The Fundamental Limitations of Agentic AI Beyond specific security vulnerabilities, experts identify deeper limitations in current AI agent technology. Symons highlighted the critical thinking gap: “If you think about human higher-level thinking, that’s one thing that maybe these models can’t really do. They can simulate it, but they can’t actually do it.” This limitation manifests in several key areas: Critical reasoning: AI agents lack genuine understanding and contextual judgment Security implementation: Current guardrails rely on natural language instructions rather than robust technical controls Autonomy limitations: Agents require significant human oversight and intervention Scalability challenges: Security vulnerabilities increase exponentially with deployment scale Industry Recommendations and Current Status Given the identified vulnerabilities, security experts offer cautious recommendations for OpenClaw implementation. Hammond stated plainly: “Speaking frankly, I would realistically tell any normal layman, don’t use it right now.” This recommendation stems from the fundamental tension between functionality and security in current agentic AI systems. The industry faces a critical challenge: for agentic AI to deliver promised productivity gains, systems must overcome inherent security vulnerabilities. Current implementations struggle to balance accessibility with protection, particularly against sophisticated prompt injection attacks that exploit the natural language processing capabilities that make these systems useful. Broader Implications for AI Development The OpenClaw experience provides valuable lessons for the broader AI industry. First, rapid viral adoption often outpaces security considerations, creating systemic vulnerabilities. Second, the distinction between genuine innovation and repackaged existing technology requires careful evaluation. Third, public perception of AI capabilities frequently exceeds current technical realities. These insights come at a crucial moment in AI development, as companies race to implement agentic systems for competitive advantage. The Moltbook incident serves as a cautionary tale about prioritizing security fundamentals before scaling experimental technologies. Conclusion OpenClaw represents both the promise and peril of current AI agent technology. While the framework demonstrates impressive integration capabilities and user-friendly design, fundamental security vulnerabilities and technical limitations undermine its practical utility. The Moltbook incident revealed how quickly experimental systems can develop critical security flaws when deployed without adequate safeguards. AI experts consistently emphasize that OpenClaw combines existing components rather than creating novel breakthroughs. More importantly, the system’s vulnerability to prompt injection attacks and authentication failures highlights the broader challenges facing agentic AI development. As the industry progresses, balancing innovation with security will remain essential for realizing AI’s transformative potential while protecting users and systems from emerging threats. FAQs Q1: What exactly is OpenClaw and why did it become so popular? OpenClaw is an open-source AI agent framework that enables users to create customizable agents communicating through natural language across messaging platforms. It gained popularity through GitHub visibility and its user-friendly approach to agent creation, despite lacking fundamental security measures. Q2: What security vulnerabilities were discovered in OpenClaw and Moltbook? Researchers found unsecured credentials in Moltbook’s database, allowing token theft and agent impersonation. The systems lacked authentication guardrails, rate limits, and protection against prompt injection attacks that could compromise sensitive data and systems. Q3: How do prompt injection attacks work against AI agents? Prompt injection involves tricking AI agents through carefully crafted inputs to perform unauthorized actions. Attackers might embed malicious instructions in emails, posts, or other inputs that agents process, potentially leading to credential theft, financial transactions, or system compromises. Q4: Are AI experts recommending against using OpenClaw currently? Yes, multiple security experts explicitly recommend against using OpenClaw in production environments due to unresolved vulnerabilities. They advise waiting for more secure implementations before deploying agentic AI systems for sensitive or critical applications. Q5: What broader lessons does the OpenClaw experience offer for AI development? The incident highlights the importance of prioritizing security fundamentals before scaling experimental technologies. It demonstrates how viral adoption can outpace safety considerations and emphasizes the need for rigorous testing of AI systems before widespread deployment. This post OpenClaw AI Exposed: The Alarming Security Flaws Behind the Hype first appeared on BitcoinWorld .
14 Feb 2026, 23:02
PancakeSwap V2 OCA/USDC pool on BSC drained of $422K
The PancakeSwap V2 pool for OCAUSDC on BSC was exploited in a suspicious transaction detected today. The attack resulted in the loss of almost $500,000 worth of USDC market, drained in a single transaction. According to reports from Blockchain security platforms, the attacker exploited a vulnerability in the deflationary sellOCA() logic, giving them access to manipulate the pool’s reserves. The final amount the attacker got away with was reportedly approximately $422,000. The exploit involved the use of flash loans and flash swaps combined with repeated calls to OCA’s swapHelper function. This removed OCA tokens directly from the liquidity pool during swaps, artificially inflating the on-pair price of OCA and enabling the drainage of USDC How did the OCA/USDC exploit happen? The attack was reportedly executed via three transactions. The first to carry out the exploit, and the following two to serve as additional builder bribes. “In total, 43 BNB plus 69 BNB were paid to 48club-puissant-builder, leaving an estimated final profit of $340K,” Blocksec Phalcon wrote on X about the incident, adding that another transaction in the same block also failed at position 52, likely because it was frontrun by the attacker. Flash loans on PancakeSwap allow users to borrow significant amounts of crypto assets without collateral; however, the borrowed amount plus fees must be repaid within the same transaction block. They are primarily used in arbitrage and liquidation strategies on the Binance Smart Chain, and the loans are usually facilitated by PancakeSwap V3’s flash swap function. Another flash loan attack was detected weeks ago In December 2025, an exploit allowed an attacker to withdraw approximately 138.6 WBNB from the PancakeSwap liquidity pool for the DMi/WBNB pair, netting approximately $120,000. That attack demonstrated how a combination of flash loans and manipulation of the AMM pair’s internal reserves via sync() and callback functions is capable of being used to completely deplete the pool. The attacker first created the exploit contract and called the f0ded652() function, a specialized entry point into the contract, after which the contract then calls flashLoan from the Moolah protocol, requesting approximately 102,693 WBNB. Upon receiving the flash loan, the contract initiates the onMoolahFlashLoan(…) callback. The first thing the callback does is find out the DMi token balance in the PancakeSwap pool in order to prepare for the pair’s reserve manipulation. It should be noted that the vulnerability is not in the flash loan, but in the PancakeSwap contract, allowing manipulation of reserves via a combination of flash swap and sync() without protection against malicious callbacks. Get 8% CASHBACK when you spend crypto with COCA Visa card. Order your FREE card.
