MiniMax published cryptoanalysis results obtained using an internal model
Read more
Pandaily
pandaily.com

MiniMax published cryptoanalysis results obtained using an internal model

MiniMax has disclosed the results of automated cryptoanalysis experiments conducted using its proprietary research system, which it named MiniMax's Internal Model, or MIM. A company overview, published in its English blog and public repository on November 20th, details the findings concerning academic developments, candidates from historical competitions, and examples of public challenges.

The results were categorized into three main areas: analysis of past NIST competition candidates, study of peer-reviewed cryptographic constructions, and the development of faster solvers for mathematical problems. The company emphasizes that these categories should not be interpreted as proof that the artificial intelligence system has broken encryption used in everyday communication.

MiniMax explicitly states that this work does not affect current cryptographic standards. Historical candidates submitted to the standardization competition are not equivalent to the finally selected standards, and public challenge examples test specific parameters rather than every implementation of the corresponding cryptosystem.

Among the reported results, MiniMax states that MIM solved 94 tasks within the TII McEliece challenges, including 55 that the company considers previously unsolved. Furthermore, improvements to the analysis of HAWK parameters were presented, and weaknesses were discovered in some early designs. This data represents the company's own conclusions; the number of solved tasks is not evidence of recovering keys from 94 active services or compromising an entire family of standardized algorithms.

The company presents these experiments as a test of the ability to maintain mathematical reasoning combined with implementation work. While it claims practical results and increased speed compared to previous analyses, such comparisons depend on the chosen goals, parameters, hardware, and benchmark methods. Pandaily did not conduct independent reproduction of the experiments or verify the claimed performance advantages.

Regarding disclosure, MiniMax reported that it preliminarily shared the results with the authors of the relevant schemes. Its repository overview lists technical reports and conceptual proof artifacts released on November 20th. However, the notification of preliminary publication is a company statement, not confirmation that all authors have approved the findings or completed necessary corrective actions.

The MIM model itself is not a public release; the company provides access for testing upon request to qualified researchers. Therefore, public research artifacts must differ from open model weights or a public product. Pandaily reviewed only the announcement and the repository overview, without downloading or executing exploit code, reproducing attacks, or performing operational key recovery. Independent cryptographic analysis remains necessary to establish the scope and significance of the claims.

Similar stories

Simplex AI releases Linger code for local personal agents
Read more
pandaily.com

Simplex AI releases Linger code for local personal agents

Simplex AI has made the source code for Linger public—a personal agent designed to run on a user's own computer using local language models. The project was featured in Chinese publications on November 20th. The project founder, Tony Zhang, previously led the DeepSeek-VL project.

Linger is a product of Simplex AI, not a release from DeepSeek. The system's central agent, named Another Me, is designed to structure information from emails, calendars, chats, and meetings into a unified user knowledge base. It can also identify tasks requiring follow-up and delegate them to other agents. The project description also mentions browser and desktop operation capabilities with the option for human intervention when necessary. It should be noted that these functions are described by the developers and have not undergone independent product testing.

The local inference component, called PolyStrata, distributes the loads of 'mixture-of-experts' type models across the Graphics Processing Unit (GPU) memory, System Random Access Memory (RAM), and SSD storage, instead of requiring all experts to reside on the video card. More frequently used experts remain closer to the GPU, while less demanded parts can be served through other memory levels.

Simplex AI reports that the Qwen3.8-Flash-Next model with 125 billion parameters generates 45.5 tokens per second on an RTX 3060 with 12GB of VRAM and 64GB of RAM, and 179 tokens per second on an RTX 5090 with 32GB. These figures are manufacturer-reported decoding speed measurements and do not represent full agent task time characteristics or guarantees of equivalent quality to cloud models. Additional hardware planning tables include budget memory configuration simulations, which limits comparisons between different machines.

The public release of the code does not imply unlimited licensing. The main license modifies Apache 2.0 by adding additional conditions: running Linger as a hosted or managed service for third parties requires written permission from the manufacturer, and removing or changing the frontend branding is only possible with a commercial license. The terms allow for internal commercial use and use as a backend for the organization's own applications.

At the time of checking, the main website indicated that downloads for macOS, Windows, and Linux as ready-made packages are expected in the future, despite the fact that the source code and setup documentation were publicly available. Code availability should not be confused with the release of a finished installer. Furthermore, local inference does not guarantee fully autonomous operation, as connected applications, voice commands, remote models, and background tasks using external subscriptions may transmit data to external providers. The security documentation describes server assertions and checks, not a sandbox isolating agents from owner files. Pandaily did not install the software nor reproduce the claimed performance.

Huawei introduces Kirin 9050 security architecture with post-quantum cryptography support
Read more
technode.com

Huawei introduces Kirin 9050 security architecture with post-quantum cryptography support

Huawei revealed details of the Kirin 9050 chip series security architecture on Wednesday. It was stated that the mobile chips support post-quantum cryptography and have passed testing for resistance to quantum attacks.

According to information from Huawei, this architecture combines a mobile security processor with an integrated secure element chip and has received CCRC EAL5+ security certification.

These two security components function together to ensure data protection, confidentiality, as well as communication and system secure boot security. Post-quantum cryptography is designed to protect information from potential future quantum computers that could compromise widely used encryption methods.

This disclosure followed recent technical reviews by Huawei concerning the computational, graphical, and display capabilities of the Kirin 9050.

Cybersecurity company warns of fraud after failures recorded at Microsoft
Read more
rtp.pt

Cybersecurity company warns of fraud after failures recorded at Microsoft

A cybersecurity company issued an alert regarding the increase in fraud that followed operational failures at Microsoft. As a result of these incidents, the Cybersecurity Center requested that all users exercise increased attention and caution.

Regarding the impact of these failures, it was reported that there was a collapse at Microsoft affecting the Health sector. However, the Government assured that the functioning of the State remained normal during the period of technical problems.

Popular