CVE-2026-33697
HighSummary
Cocos AI to system zaufanego obliczania dla AI, który ma podatność na atak relay w implementacji attested TLS (aTLS) w wersjach od v0.4.0 do v0.8.2. Atakujący może wyciągnąć ephemeryczny klucz prywatny TLS używany podczas atestacji, co pozwala na przejęcie sesji TLS.
Risk Assessment
Podatność ta może prowadzić do fałszywej autoryzacji, umożliwiając atakującemu podszywanie się pod usługi CoCoS i dostęp do danych lub operacji, co stanowi poważne zagrożenie dla bezpieczeństwa organizacji.
Recommendation
Zaleca się aktualizację systemu CoCoS do najnowszej wersji, aby usunąć tę podatność oraz wdrożenie dodatkowych środków zabezpieczających, aby zminimalizować ryzyko ataków relay.
Related vulnerabilities
- CVE-2026-90898Critical
Bifrost registers MCP clients through its management API, where a stdio client is a command plus args that is started immediately upon client addition, with no MCP handshake required. Authentication is disabled by default (governance.auth_config.is_enabled=false), meaning every caller is treated as a local admin. A single unauthenticated POST /api/mcp/client is enough to run a program as the Bifrost process user (appuser on the official image).
- CVE-2026-90703Critical
A vulnerability has been found in D-Link DWR-M921 1.1.52 in the system function of the file /boafrm/formDiskCreateShare. Manipulation of the folderpath argument leads to OS command injection. The attack can be launched remotely, and the exploit has been publicly disclosed and may be used.
- CVE-2026-90702Critical
A flaw has been found in D-Link DWR-M921 1.1.52 in the system function of the file /boafrm/formDiskFormat. Manipulation of the partition argument causes OS command injection. The attack may be initiated remotely, and the exploit has been published and may be used.
- CVE-2026-90699Critical
A weakness has been identified in D-Link DWR-M920 1.1.7 affecting the sub_41E60C function of the file /boafrm/formPinManageSetup. Manipulation of the newPin argument causes OS command injection. The attack can be initiated remotely, and the exploit has been made available to the public and could be used for attacks.
- CVE-2026-90693Critical
A flaw has been found in D-Link DIR-878 120B05 impacting the SetWan3Settings function of the WAN Settings component. Manipulation of the Primary/Secondary argument causes a stack-based buffer overflow. Remote exploitation of the attack is possible.
- CVE-2026-90692Critical
A vulnerability was detected in D-Link DIR-878 120B05 affecting the SetDynamicDNSIPv6Settings function of the Dynamic DNS IPv6 Settings component. Manipulation of the IPv6Address/Hostname argument results in a stack-based buffer overflow. The attack may be launched remotely.
- CVE-2026-85192Critical
The Joomla Conditional Content extension (regularlabs.com) in versions < 8.0.0 allows authenticated, privileged remote code execution. Conditional Content Pro accepts inline PHP Condition Rules in article syntax, and in affected versions the PHP is passed to the Conditions evaluator without checking who authored the article. Joomla's normal Author text filter preserves the syntax, so publishing the article causes the code to run as the web-server process.
- CVE-2026-82787Critical
A missing authentication for critical function vulnerability exists in CPSL-08P1EN. If this vulnerability is exploited, an affected product may be operated by a remote attacker without authentication.
- CVE-2026-90680Critical
A stack-based buffer overflow vulnerability was discovered in D-Link DIR-823G version 1.0.2B05_20181207, in the strcpy function of /HNAP1/SetStaticRouteSettings (HNAP1 component). Manipulating the PAddress/SubnetMask/Gateway arguments triggers the overflow. The attack can be launched remotely.
- CVE-2026-90608Critical
A buffer overflow vulnerability was found in Totolink A3002MU version Hh-B20211125.1046, in the formPortFw function of /boafrm/formPortFw (boa component). Manipulating the service_type argument causes the overflow. The attack can be initiated remotely and the exploit has been published.
Original NVD description (English source)
Cocos AI is a confidential computing system for AI. The current implementation of attested TLS (aTLS) in CoCoS is vulnerable to a relay attack affecting all versions from v0.4.0 through v0.8.2. This vulnerability is present in both the AMD SEV-SNP and Intel TDX deployment targets supported by CoCoS. In the affected design, an attacker may be able to extract the ephemeral TLS private key used during the intra-handshake attestation. Because the attestation evidence is bound to the ephemeral key but not to the TLS channel, possession of that key is sufficient to relay or divert the attested TLS session. A client will accept the connection under false assumptions about the endpoint it is communicating with — the attestation report cannot distinguish the genuine attested service from the attacker's relay. This undermines the intended authentication guarantees of attested TLS. A successful attack may allow an attacker to impersonate an attested CoCoS service and access data or operations tha

