Scan to download
BTC $76,910.19 -1.51%
ETH $2,118.04 -3.22%
BNB $639.78 -2.16%
XRP $1.38 -2.15%
SOL $84.64 -2.46%
TRX $0.3549 +0.06%
DOGE $0.1041 -5.80%
ADA $0.2489 -2.75%
BCH $365.01 -11.99%
LINK $9.46 -3.26%
HYPE $45.66 +6.54%
AAVE $88.28 -3.10%
SUI $1.03 -4.07%
XLM $0.1471 -3.01%
ZEC $529.54 +1.74%
BTC $76,910.19 -1.51%
ETH $2,118.04 -3.22%
BNB $639.78 -2.16%
XRP $1.38 -2.15%
SOL $84.64 -2.46%
TRX $0.3549 +0.06%
DOGE $0.1041 -5.80%
ADA $0.2489 -2.75%
BCH $365.01 -11.99%
LINK $9.46 -3.26%
HYPE $45.66 +6.54%
AAVE $88.28 -3.10%
SUI $1.03 -4.07%
XLM $0.1471 -3.01%
ZEC $529.54 +1.74%

pol

AI Agent Security Risk Exposure: Attackers Can Exploit "Memory Pollution" to Induce Misoperation of Funds

The GoPlus Security team has disclosed a new type of attack in its AgentGuard AI project: inducing AI agents to perform unauthorized sensitive operations through "memory poisoning." This attack method does not rely on traditional vulnerabilities or malicious code but exploits the long-term memory mechanism of AI agents. For example, an attacker first induces the agent to "remember preferences," such as "usually prioritizing proactive refunds instead of waiting for chargebacks," and then uses vague expressions like "process as usual" or "execute as before" in subsequent instructions, thereby triggering automated financial operations.GoPlus points out that the key risk in such cases lies in the AI agent mistakenly treating "historical preferences" as a basis for authorization, leading to financial losses or security incidents in operations such as refunds, transfers, and configuration changes. To address this issue, the team has proposed several protective recommendations, including:Operations involving refunds, transfers, deletions, or sensitive configurations must require explicit confirmation in the current session.Memory-related instructions like "habit," "usual way," and "as before" should be regarded as high-risk state changes.Long-term memory must have a traceability mechanism (writer, time, confirmation status).Vague instructions should automatically elevate the risk level and trigger secondary verification.Long-term memory must not replace real-time authorization processes.The team emphasizes that the "AI agent memory system" should be viewed as a potential attack surface and should be constrained and audited through a dedicated security framework.

The Polish Parliament is reviewing four cryptocurrency bills simultaneously

The Speaker of the Polish Sejm, Włodzimierz Czarzasty, announced that the parliament has officially begun reviewing four competing regulatory bills for crypto assets, following President Karol Nawrocki's veto of related legislation twice. This review involves multiple legislative proposals from the government, the presidential office, the Poland 2050 party, and the Confederation party, with a second reading vote expected to take place on Thursday. The core disagreement centers on the scope of the Polish Financial Supervision Authority (KNF) regarding account freezing powers and the maximum penalties for violations. The presidential draft sets the maximum fine at approximately 20 million zlotys (about 5.5 million USD), while the Ministry of Finance's version raises it to 25 million zlotys (about 6.9 million USD).Meanwhile, the opposition Law and Justice party (PiS), after withdrawing support for earlier regulatory proposals, submitted a separate bill on Monday advocating for a complete ban on crypto asset-related activities in Poland, further complicating the regulatory discussion. Speaker Czarzasty stated that the PiS ban draft will enter the review process only after the four main regulatory bills are completed and questioned the connection between crypto industry funding and political activities, specifically inquiring about potential political financing issues, including those involving zondacrypto.
app_icon
ChainCatcher Building the Web3 world with innovations.