Google, OpenAI, Microsoft ແລະ Anthropic ຮັບຮອງກົດດຽວ ເພື່ອປະຕິເສດການເກັບຂໍ້ມູນຝຶກ AI
Google, OpenAI, Microsoft ແລະ Anthropic ໄດ້ໃຫ້ຄຳໝັ້ນວ່າ ຈະເຄົາລົບຄຳສັ່ງໃໝ່ໃນໄຟລ໌ robots.txt ທີ່ໃຫ້ເຈົ້າຂອງເວັບໄຊກຳນົດວ່າ ເນື້ອຫາຂອງຕົນສາມາດຖືກເກັບໄປຝຶກ generative AI ໄດ້ ຫລື ບໍ່, ອີງຕາມ Engadget ທີ່ລາຍງານໃນວັນທີ 31 ຕຸລາ. ກົດນີ້ມາຈາກຂໍ້ສະເໜີທີ່ກຳລັງພິຈາລະນາຢູ່ IETF ເຊິ່ງເປັນອົງກອນມາດຕະຖານອິນເຕີເນັດ, ຂຽນໂດຍທ່ານ Kevin Kelley ຫົວໜ້າດ້ານເຕັກນິກມາດຕະຖານຂອງ Anthropic.
ປັດຈຸບັນ ເຈົ້າຂອງເວັບໄຊຕ້ອງລະບຸ crawler ຂອງແຕ່ລະບໍລິສັດ ແລະ ຕັ້ງກົດແຍກກັນສຳລັບແຕ່ລະບໍລິສັດ. ຕາມຂໍ້ສະເໜີນີ້, user agent ໃໝ່ພຽງອັນດຽວສຳລັບການຝຶກ AI ຄື “__usage_ai-training” ຈະບລັອກ crawler ທີ່ເກັບຂໍ້ມູນໄປຝຶກ AI, ໂດຍສາມາດເພີ່ມແຖວ “Allow” ສຳລັບບາງສ່ວນຂອງເວັບໄຊ ເຊັ່ນ ຊຸດຂໍ້ມູນສຳລັບສື່ມວນຊົນ. ກົດນີ້ບໍ່ໄດ້ຢຸດການ crawl ທົ່ວໄປ ຫລື ເອົາເວັບໄຊອອກຈາກ Google Search; Google ຈະຖືວ່າເປັນສັນຍານຢຸດ ສຳລັບ crawler “Google-Extended” ທີ່ເກັບຂໍ້ມູນໄປຝຶກແບບຈຳລອງ Gemini.
ຄຳສັ່ງນີ້ບໍ່ສາມາດຍົກເລີກການເກັບຂໍ້ມູນທີ່ເກີດຂຶ້ນແລ້ວໄດ້. ອີງຕາມ Engadget, Anthropic ຮອງຮັບແລ້ວ, Microsoft ຈະເລີ່ມຮອງຮັບໃນວັນທີ 1 ກຸມພາ 2027, Google ມີແຜນຮອງຮັບໃນຕົ້ນປີ 2027, ແລະ OpenAI ກ່າວວ່າ ຈະນຳໃຊ້ໃນອີກບໍ່ເທົ່າໃດເດືອນຂ້າງໜ້າ.
Google, OpenAI, Microsoft and Anthropic Back One AI-Training Opt-Out
Google, OpenAI, Microsoft and Anthropic have committed to honouring a new robots.txt directive that lets website owners say whether their content may be collected to train generative AI, Engadget reported on 31 October. The rule comes from a proposal under consideration at the IETF, the internet standards body, written by Kevin Kelley, Anthropic’s technical lead for standards.
Today, site owners have to identify each company’s crawler and add a separate rule for each one. Under the proposal, a single new user agent for AI training, “__usage_ai-training”, blocks crawlers that collect data for AI training, with optional “Allow” lines for specific sections such as a press kit. It does not stop normal crawling or remove a site from Google Search; Google will treat it as a stop signal for its “Google-Extended” crawler, which gathers material for training Gemini models.
The directive cannot undo past scraping. Anthropic already supports it, Microsoft will start on 1 February 2027, Google plans support in early 2027, and OpenAI said it will implement it in the coming months, according to Engadget.
Source: Engadget – https://www.engadget.com/2303225/google-openai-anthropic-ai-crawling-controls/
