September 2nd, CentralInternet Information OfficeToday's release of "Chirant" AI ApplicationsUpdates on progress and related results of the second phase of the Operation。
THE CAMPAIGN FOCUSED ON ISSUES SUCH AS THE DISSEMINATION OF FALSE INFORMATION, THE DISSEMINATION OF VIOLENT AND TACKY CONTENT, THE COUNTERFEITING OF OTHERS AND THE ABUSE OF THE RIGHTS AND INTERESTS OF MINORS。
According to the official report, to date, the operation has accumulated over 5.61 million violations of the law, checked accounts with more than 49,000 and disposed of over 2,400 sites, applications, etc. At the same time, a range of targeted governance measures have been introduced by the local network offices:
- (b) The Beijing UNOI guidance platform upgrades its vetting strategy to improve its ability to identify and dispose of new variants of non-compliance
- SHANGHAI HAS URGED THE PLATFORM TO SELF-CENSOR AND LAUNCH AI TO DIRECT THE PUBLIC TO THE IDENTIFICATION OF RISKS USING A “KNOWLEDGE PACKAGE”
- THE ZHEJIANG OFFICE OF LETTERS AND LETTERS PROVIDES "ONE-SIZE-FITS-ALL" GUIDANCE ON COMPLIANCE WITH KEY TERRITORIAL PLATFORMS AND INVESTIGATES THE MOST PROBLEMATIC AI APPLICATIONS IN ACCORDANCE WITH THE LAW
- An example of a focused platform organized by the Guangdong Network Office will provide a point-of-custody review of multiple applications。
In addition, platforms such as tremors, fast hands, microblogging, tampons, hundreds of degrees, bells, small red books, bean petals and poachers further enhance technical empowerment, optimize model multi-modular identification models, dynamically expand features such as human faces, acoustic textures and violation samples, real-time face management systems for top-liners, improve the efficiency of intelligence detection and interception of non-compliance and cumulatively issue various governance bulletins。
Platforms such as bean buns, Yuanbao, Chiu Quan and Man Quan have strictly closed the original language content, reinforced AI ' s requirement to produce synthetic content markers, and prevented the transmission of illegal information from the source. The Chinese, Mi, OPPO, vivo and other app stores establish strict developer access norms and upscaling requirements, carry out spot-checks and check-ups on the App and effectively intercept violations。
THE OFFICE HAS ALSO PUBLISHED A NUMBER OF TYPICAL CASES OF NON-COMPLIANCE DETECTED DURING THE CAMPAIGN,1 WHICH ARE SUMMARIZED BELOW:
- SOME OF THE PLATFORM'S ACCOUNTS, SUCH AS THOSE OF WEIBO, WIAN, QUIZ AND WISDOM, USE CLASSICS SUCH AS AL'AI'S “MAGIC” THE THREE NATIONS, THE WEST TRAVEL, TO SUBVERT THE CREATION OF IMPORTANT FIGURES AND TO FABRICATE LOW-QUALITY HUNTING DRAMAS
- SOME ACCOUNTS USE AI TO GENERATE SYNTHETIC FALSE DISASTER SCENES, RESCUE IMAGES, FABRICATE FALSE SOCIAL HOTSPOTS, OR CREATE AN AI DIGITAL PERSON TO CARRY FALSE PROPAGANDA TO MISLEAD THE ELDERLY
- SOME OF THE ACCOUNTS OF THE PLATFORMS, SUCH AS TREMORS, QUICK HANDS, LITTLE RED BOOKS, MICRO-LETTERS, ETC., PROFITED FROM THE USE OF THE AI "SMOKE" "SMOKE" TECHNOLOGY TO IMPERSONATE THE IMAGE OF PUBLIC FIGURES AND TO GENERATE VIDEO MACHINES FOR THE SALE OF IMAGES SUCH AS LOW-LIFE DANCES, HUGS AND KISSES
- A FEW DAYS AGO, A GROUP OF ACCOUNTS FROM THE PLATFORMS, SUCH AS 哩 哔 ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;
- (A) THE PUBLICATION OF A PART OF THE ACCOUNT NUMBER OF A SMALL RED BOOK, MICROBLOGGING, TREMORS, QUICK-HANDS, ETC., ON THE CONTENT PRODUCED BY AI ON THE PREGNANCY OF MINORS, FIGHTING, ETC., WHICH IS NOT IN KEEPING WITH THEIR AGE CHARACTERISTICS, AND THE PRODUCTION OF A SERIES OF VIDEOS THAT SPREAD “TOXIOUS ANIMATIONS”, “TORNURE” OF CHILDREN, “FALSE” OF CHILDREN, HARMING AND DEFUNCT IMAGES OF MINORS
- SOME OF THE PLATFORM ACCOUNTS, SUCH AS TWITTER, MICROBLOGGING AND SHIVERING, ARE AI “HOSTING” SOFTWARE STREAMS, SIMULATE REAL-PERSON OPERATING ACCOUNTS, COMMENT INTERACTIONS AND GENERATE A LARGE NUMBER OF LOW-QUALITY HOMOGENEIC CASES FOR WEB-BASED NAVAL ACTIVITIES。
- In addition, the “Nt” “Fashioned” “Sweet * Circle” “Too **o” web site features, profiles and dialogues containing pornographic and other information that is harmful to the physical and mental health of minors, are subject to legal scrutiny and are followed by procedural measures such as warnings, fines, layoffs or closures。
AMONG THEM, THE ORIGINAL TERM "MAGIC MODIFICATION" REFERS TO A SECOND CREATION OF A CLASSIC VIDEO WORK, BUT SOME OF THE CREATORS USE AI TECHNOLOGY TO CREATE A BOTTOM-UP ADAPTATION OF THEIR FAMOUS WORK, CREATING LOW-LIFE, GROTESQUE CONTENT TO CAPTURE TRAFFICThe Web Office calls this "digital water".
“TOXIC ANIMATION” MEANS THE FORM OF ANIMATION OF A CHILD ON THE SURFACE, WHILE AI, WHICH CONTAINS NEGATIVE ELEMENTS SUCH AS VIOLENCE, TERROR OR SOFT PORNOGRAPHY, PRODUCES VIDEO, WHICH POSES A POTENTIAL RISK TO THE PHYSICAL AND MENTAL HEALTH OF MINORS。
THE INITIATIVE, WHICH IS AIMED AT THE SYSTEMATIC MANAGEMENT OF AI ' S SYNTHETIC CONTENT DISORDER, HAS EFFECTIVELY CURBED THE SPREAD OF RELATED VIOLATIONS. IN THE NEXT STEP, ON THE BASIS OF A REVIEW OF THE EFFECTIVENESS OF GOVERNANCE, THE WEB SECTION WILL FOLLOW CLOSELY THE NEW RISKS OF AI GENERATING NEW INFORMATION ON SYNTHETIC CONTENT AND TAKE TIMELY MEASURES TO PREVENT MISUSE OF AI TECHNOLOGY。