Shulou(Shulou.com)11/24 Report--

On July 7, 2023 World Artificial Intelligence Conference (WAIC)"Focus·Big Model Era AIGC New Wave-Trusted AI" Forum was held, Ant Group jointly with Tsinghua University released AI security detection platform "Ant Discrimination 2.0." Ant Discrimination 2.0 can realize the detection of generative AI model with generative AI capability, identify hundreds of risks in three categories of data security, content security and science and technology ethics, cover various data and task types such as tables, texts and images, and is the first AI security detection platform in the industry to realize industry-level application. It is understood that this product has also been selected as the "treasure of the town hall" of this conference.

(Picture: Ant Group and Tsinghua University jointly release AI safety detection platform "Ant Discrimination 2.0")

AIGC security has become an important issue in the current development of artificial intelligence research. He Jifeng, academician of Chinese Academy of Sciences, delivered a keynote speech on "Trusted Artificial Intelligence Outlook". He believes that "AI can no longer be seen as a simple technical job, but as an engine of change for the entire organization." Trusted artificial intelligence should be committed to ensuring data security and credibility, system behavior can be held accountable, algorithm model can be explained, network environment can be trusted, legal ethics can be trusted and other issues. "

At the same time, Ant Group announced that "Ant Discrimination 2.0" is fully open, providing AIGC security, AI interpretability and AI robustness three Detection Tools free of charge to developers around the world, which can serve large-scale complex business scenarios in digital finance, education, culture, medical treatment, e-commerce and other fields.

It is understood that "ant discrimination 2.0" two hard core capabilities, one is to realize the use of generative capability detection generative model. The detection standard covers three categories: content security, data security and ethical security. It can complete risk confrontation detection including hundreds of dimensions such as personal privacy, ideology, illegal crime, prejudice and discrimination for the generative content of large model, and generate detection report to help the large model to be optimized continuously and more pertinently.

Li Junkui, president of Ant Group's large security business group technology department, said,"The generative large model is a kind of 'deep black box' technology.'Ant Discrimination 2.0' can realize the inductive detection calculation of the 'generative robot' on the 'AIGC generative model' through intelligent game confrontation technology, simulation of black production and automatic generation of massive test sets, much like a 24-hour sleepless 'security hacker' looking for a large model. So we can find the weakness and security problem of the large model. This kind of AI evaluation of AI is also teaching and learning. For example, the generative examiner assesses the generative athletes, and the abilities of both sides improve each other in the confrontation, further improving the AI safety level of the whole industry. "

Another hardcore capability is that Ant Discrimination 2.0 incorporates interpretability Detection Tools. Comprehensive AI technology and expert prior knowledge, through visualization, logical reasoning, causal inference and other technologies, quantitative analysis of AI system interpretation quality from 7 dimensions such as completeness, accuracy and stability and more than 20 evaluation indicators, helping users to verify and optimize interpretable schemes more clearly.

It is understood that "ant discrimination 2.0" integrates ant nearly 10 years of trusted AI practice, more than 1000 trusted AI patents, and supports zero-code evaluation. Developers can identify and mine model vulnerabilities with one click by connecting AI models to ant authentication through API interfaces.

Ant Group believes that the deep application of artificial intelligence not only puts forward higher requirements for data, algorithms and computing power, but also poses more challenges to security, privacy and ethics. Only on the premise of ensuring data security and privacy protection and improving artificial intelligence ethics and security can AI technology truly release its application value.

At this forum, the AIGC Trusted Initiative was jointly launched by China Academy of Information Technology, Shanghai Artificial Intelligence Laboratory, Wuhan University, Ant Group and other units. The initiative has been participated by dozens of units. Focusing on the economic, security, privacy and data governance issues that may be caused by artificial intelligence, it proposes to build a global governance cooperation framework for the trusted development of AIGC, and comprehensively enhance the trusted engineering capability of AIGC by adopting secure and credible data resources, technical frameworks, computing methods and software platforms, so as to ensure the security, transparency and releasability of generative AI to the greatest extent.

(Photo: At the 2023 WAIC Trusted AI Forum, the "industry-university-research and application" community jointly launched the AIGC Trusted Initiative)

The forum also released the White Paper on Trusted AI Technology and Application Progress (2023) jointly compiled by China Academy of Information Technology, Tsinghua University and Ant Group. Based on the unified cognitive dimension of trusted AI, the report sorts out and summarizes the current development status of trusted AI, improves the new requirements caused by large models and AIGC facing the new stage, forms a trusted AI evaluation system and tools guaranteed by technology, and explores the development of a new generation of trusted artificial intelligence.

According to public information, ants started the practice and exploration of trusted AI since 2015, which is one of the key layouts of ant group in the field of artificial intelligence. Through the breakthrough of trusted AI technology, Ant Group has built a set of world-leading intelligent risk control solutions; Ant AI security detection platform has successively won the recognition of excellent achievements for the third anniversary of the construction of Shanghai Financial Science and Technology Center in 2022 and the benchmark cases of trusted artificial intelligence practice of ICT.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.