On May 12th, there's been a "one key to generate a bomb" on the Internet recentlyScrawl notes"THE AI WRITING TOOL, WHICH PRODUCES FALSE CONSUMPTION EXPERIENCES AND COMMODITY RECOMMENDATIONS AT VERY LOW COST。
IN RECENT DAYS, THE INTERMEDIATE PEOPLE'S COURT OF HANGZHOU CITY, ZHEJIANG PROVINCE, HAS REVEALED A TYPICAL CASE INVOLVING A SOCIAL PLATFORM, THE FIRST CASE OF UNFAIR COMPETITION INVOLVING AI'S “GRAIN NOTE”, IN WHICH A REASONABLE DUTY OF CARE WAS ESTABLISHED FOR THE GENERATION OF ARTIFICIAL INTELLIGENCE SERVICE PROVIDERS。
THE HANGZHOU COURT'S FINAL DECISION MADE IT CLEAR THAT THE AI TOOL WAS DIRECTED TO CREATE FALSE CONSUMPTION-SHARING, UNDERMINE THE REAL CONTENT ECOLOGY OF THE PLATFORM, CONSTITUTE AN UNDUE COMPETITION, AND PROVIDE A LEGAL YARDSTICK FOR THE MANAGEMENT OF AI CONTENT AND “DIGITAL WATER” MANAGEMENT。
IN THIS CASE, THE AI WRITING TOOL OPERATED BY THE TWO COMPANIES, WHICH PROVIDES TARGETED GENERATION SERVICES SUCH AS “BREEDING” “WRITTEN HEADLINE” FOR SOCIAL PLATFORMS, INDUCES USERS TO PUBLISH CONTENT WITHOUT REAL EXPERIENCE AND TO CHARGE FOR PROFIT。
THE COURT FOUND THAT THE REAL CONTENT OF THE PLATFORM, WHICH HAD EVOLVED OVER A LONG PERIOD OF TIME, WAS A LEGALLY PROTECTED COMPETITIVE INTEREST; THAT THE DEFENDANT HAD FAILED TO DEFEND ITSELF AS A TECHNOLOGY-NEUTRAL DEFENCE AND THAT AI TECHNOLOGY ITSELF WAS NEUTRAL, BUT THAT COMMERCIAL APPLICATIONS WERE SUBJECT TO THE PLATFORM ' S RULES AND LEGAL BOTTOM LINES。
THE COURT ' S INNOVATION USES A FOUR-FACTOR DETERMINATION THAT THE CONDUCT IS UNLAWFUL: WHETHER IT IS A GENERATING AI SERVICE, WHETHER IT IS BASED ON A SPECIFIC PLATFORM, WHETHER IT IS ATTRACTIVE, WHETHER IT IS PROFITABLE, OR NOTThe defendant was finally sentenced to cease service and to pay $100,000。
FOR THE FIRST TIME, THE CASE CLARIFIED THE REASONABLE CARE OBLIGATIONS OF AI SERVICE PROVIDERS, REQUIRING THE PLATFORM, MODEL MANUFACTURERS AND SERVICE PROVIDERS TO WORK TOGETHER TO STRENGTHEN CONTENT LABELLING, RISK ALERTS AND COMPLIANCE CONTROLS, AND TO PROMOTE THE HEALTHY DEVELOPMENT OF AI TECHNOLOGY ON LEGITIMATE COMPLIANCE TRACKS。
According to Judge Zhang, in the future, attention will have to be paid to the development of models, the generation of content into compliance control, and to the formation of a chain-wide governance landscape。
First, to ensure that trained data and model designs are in compliance, model orientation should not be optimized to generate non-compliance elements for a specific platform。
SECOND, RISK ALERTS AND CONTENT CONTROLS SHOULD BE STRENGTHENED, WITH NOTABLE INDICATIONS OF THE LIMITATIONS OF THE CONTENT GENERATED BY AI, A CLEAR BAN ON THE PUBLICATION OF FALSE INFORMATION, AND SPECIFIC PLATFORM RULES THAT LIMIT THE SCENES OF IRREGULARITIES DIRECTED TO A GIVEN PLATFORM。
Third, content identification and user data protection should be in place。
Fourth, a complaints reporting mechanism should be established and the content of violations or abuses should be dealt with in a timely manner。