生成AIのガバナンスとこれから

2024年9月12日に開催されたCysec 10周年記念行事で発表したLT資料です。
生成AIのガバナンスについて、法規制やガイドラインが出始めていますが、それらを俯瞰した結果を現在地として示してします。
また、これから進めて行く上で役立つ考え方や動向を共有しています。

Transcript

1. Governance of Generative AI Where we are now and where we are going in the governance of generative AI. Hirokazu Yoshida / 2024.09.12 / At Tokyo Denki University
2. ੜ੒AIͷΨόφϯε ੜ੒AIͷΨόφϯεͷݱࡏ஍ͱ͜Ε͔Β ٢ాͻΖ͔ͣ / 2024.09.12 / ౦ژిػେֶʹͯ
3. Who am I !? Hirokazu Yoshida @ CloudNative Inc. Job : Security Engineer & Director Community : Security-JAWS Certi fi cation : ɹɹɹPIIP Recent work : ɹData Governance / Security / Zero Trust ɹSIEM / EDR / SASE, DLP…etc CySec : 2nd Graduates
5. Attention ! • ࣌ؒͷؔ܎্ɺੜ੒AIʹؔ࿈ͨ͠ϓϥΠόγʔʹ͍ͭͯ͸औѻ·ͤΜ • Appendixʹؔ࿈ϦϯΫ͕͋ΔͷͰɺࢀߟʹ͍ͯͩ͘͠͞ • ຊηογϣϯͷ಺༰͸ɺݸਓͷݟղͰ͋Γɺಛఆͷஂମͷओு΍ҙݟ Λࣔ͢΋ͷͰ͸͋Γ·ͤΜ
6. ಥવͰ͕͢
7. ੜ੒AI࢖ͬͯ·͔͢ʁ
8. making eye contact • ΄΅ຖ೔࢖͍ͬͯ·͢ʂ • ֶۀͰ໾ཱ͍ͯͯ·͢ʂ • ϓϥΠϕʔτͰ΋࢖͍ͬͯ·͢ʂ • ੜ੒AI͕ͳ͍ੜ׆ʹ͸΋͏໭Ε·ͤΜʂ • ͦΜͳ΋ͷʹཔΔೈऑऀͰ͸͋Γ·ͤΜʂ
9. Trend Trends in Generated AI (subjective) • ~2022/11: ChatGPTҎલ͸ઐ༻AI͕ओྲྀ ( ғޟ, ҩྍ, ੡඼඼࣭etc…) • 2023/1~6: ڪΖ͘͠൚༻తͰݡ͍gpt-3.5Λ࢖ͬͨاۀઐ༻ChatGPTߏஙϒʔϜ • 2023/7-12: اۀઐ༻ChatGPTݬ໓͔ΒͷRAGϒʔϜ ( OpenAIத৺ ) • 2024/1-6: ੜ੒AIઓࠃ࣌୅ ( SaaS౥ࡌAI, ೖྗtokenര૿, Anthropic/Googleͷ୆಄, Claud-3͕IQ100௒͑) • 2024/7-8: ௚ۙͷτϨϯυ ( Chat bot arenaͰ࠷ڧAI͕2िସΘΓ,௿ίετԽͱੑೳ ڧԽ, GPT-4oϚϧνϞʔμϧຊ֨Խ, GPT-4oϘΠεϞʔυ, ग़ྗtokenര૿)
10. ΋ͷ͍͢͝଎౓ͷਐԽ/ൃల/ల։ ҰํͰ໌Β͔ʹͳΔϦεΫ
11. Risks posed by AI • όΠΞεͷ͋Δ݁Ռ΍ࠩผతͳ݁Ռͷग़ྗ • ϑΟϧλʔόϒϧɺΤίʔνΣϯόʔݱ৅ • ଟ༷ੑͷ૕ࣦ • ෆద੾ͳݸਓ৘ใͷऔѻ͍ • ੜ໋ɾ਎ମɾࡒ࢈ͷ৵֐ • ϒϥοΫϘοΫεԽɺ൑அʹؔ͢Δઆ໌ཁٻ • ΤωϧΪʔ࢖༻ྔٴͼ؀ڥͷෛՙ • ػີ৘ใͷྲྀग़ • ѱ༻ • ϋϧγωʔγϣϯ • ِ৘ใɺޡ৘ใΛӏವΈʹ͢Δ͜ͱ • ஶ࡞ݖͱͷؔ܎ • ࢿ֨౳ͱͷؔ܎ • όΠΞεͷ࠶ੜ੒ "*ࣄۀऀΨΠυϥΠϯʢୈ൛ʣ
    IUUQTXXXNFUJHPKQQSFTTIUNM
12. ͦΜͳத AIʹର͢Δࣾձతͳૌٻ͕׆ൃʹ
13. Social appeal for AI governance • EU AI Act • AIγεςϜͷ։ൃ΍ར༻ʹؔ͢Δن੍๏Ͱ͋Γɺੈքॳͷแׅత ͳAIن੍ ( 2024/5 EUධٞձʹͯ࠷ऴঝೝ ) • 2022೥11݄Ҏ߱ͷChat GPT౳ͷੜ੒AIͷ୆಄Λड͚ͯɺ౰ॳҊ Ͱ͸໌֬ʹ͸ن੍ର৅ͱͳ͍ͬͯͳ͔ͬͨʮ൚༻໨తܕAIϞσ ϧʯʹ͍ͭͯ΋ɺಠࣗͷن੍͕ઃ͚ΒΕΔํ਑ͱͳͬͨ ਓ޻஌ೳʢ"*ʣ๏ɿධٞձ͕"*ʹؔ͢Δੈքॳͷنଇʹ࠷ऴঝೝ
    IUUQTXXXDPOTJMJVNFVSPQBFVFOQSFTTQSFTTSFMFBTFTBSUJ fi DJBMJOUFMMJHFODFBJBDUDPVODJMHJWFT fi OBMHSFFOMJHIUUPUIF fi STUXPSMEXJEFSVMFTPOBJ
14. Social appeal for AI governance • EU AI Act ( ͬ͘͟Γղઆ ) • ʮAIγεςϜʯͱʮ൚༻ੜ੒AIγεςϜϞσϧʯΛఆٛ • ʮఏڙऀ (։ൃऀ, ্ࢢऀ, ӡ༻ऀ)ʯͱʮར༻ऀʯʹରͯ͠ • ݸਓతͰඇ৬ۀతͳར༻ʹ͍ͭͯ͸ର৅֎ • 4ஈ֊ͷϦεΫͷఔ౓ʹԠͨ͡ن੍ΛఆΊͨ ( ੍ࡋنఆ͋Γ )
15. Social appeal for AI governance • EU AI Act ͷ4ஈ֊ͷϦεΫͷఔ౓ʹԠͨ͡ن੍ ( ͬ͘͟Γղઆ )
16. Social appeal for AI governance • ਓؒத৺ͷ AI ࣾձݪଇ • 2019 ೥ 5݄ʹ࠾୒͞ΕͨOECD ͷ AI קࠂҊʹج͍ͮͯɺ౷߹Πϊϕʔγϣϯઓ ུਪਐձٞ ( ಺ֳ෎ ) ͕ܾఆͨ͠ • ࣾձશମ͕ओମͱͳΓ࣮ݱ͢΂͖ AI ࣾձݪଇ͕ఆΊΒΕΔͱͱ΋ʹɺ͜ͷݪଇ Λ౿·͑ͯɺAI ͷ։ൃɾӡ༻౳ͷ౰ࣄऀͱͳΔࣄۀऀ͕ɺ֤ࣗͷ AI ͷ։ൃɾӡ ༻౳ͷ໨త΍ํ๏౳ʹԠ͡ɺ࣮ࢪ͢΂͖໨ඪʢAI ։ൃར༻ݪଇʣΛࣗΒఆΊɺ९ क͢΂͖Ͱ͋Δ 0&$%೔ຊ੓෎୅ද෦
    ʮ"*ʢਓ޻஌ೳʣʹؔ͢Δཧࣄձקࠂ͕࠾୒͞Ε·ͨ͠
    ೥݄೔ ʯ
    IUUQTXXXPFDEFNCKBQBOHPKQJUQS@KB@IUNM
17. Social appeal for AI governance • ਓؒத৺ͷ AI ࣾձݪଇ • ᶃਓؒத৺ͷݪଇ • ᶄڭҭɾϦςϥγʔͷݪଇ • ᶅϓϥΠόγʔ֬อͷݪଇ • ᶆηΩϡϦςΟ֬อͷݪଇ "*
    ݪଇ࣮ફͷͨΊͷ
    ΨόφϯεɾΨΠυϥΠϯ
    7FS
    
    IUUQTXXXNFUJHPKQTIJOHJLBJNPOP@JOGP@TFSWJDFBJ@TIBLBJ@KJTTPQEG@QEG • ᶇެਖ਼ڝ૪֬อͷݪଇ • ᶈެฏੑɺઆ໌੹೚ٴͼಁ໌ ੑͷݪଇ • ᶉΠϊϕʔγϣϯͷݪଇ
18. ཁ͸AIͷ։ൃ΍ར׆༻ʹ͍ͭͯ ૊৫ͱͯ͠ΨόφϯεΛޮ͔ͤΖ ͱ͍͏͜ͱ
19. ͦ΋ͦ΋ΨόφϯεͬͯԿʁ
20. What is governance? • ޠҙͱͯ͠͸ʮ౷࣏ɺࢧ഑ɺ؅ཧʯ • ◦◦◦ΨόφϯεͳͲɺ઀ଓ͢Δݴ༿ʹΑͬͯҙຯ߹͍͕มΘΔ • จ຺ʹΑͬͯඍົʹҙຯ͕มΘΔͷͰɺఆٛ͸ཁ֬ೝ • NIST CSF 2.0ͰʮGovernʢ౷࣏ʣʯ͕௥Ճ͞Εͨ
21. What is governance? • NIST CSF 2.0ʹ͓͚ΔʮGovernʢ౷࣏ʣʯ • ૊৫ͷഎܠͷཧղɺαΠόʔηΩϡϦςΟઓུ ͱαΠόʔηΩϡϦςΟɾαϓϥΠνΣʔϯɾϦ εΫͷཱ֬ɺ໾ׂɺ੹೚ɺݖݶɺํ਑ɺαΠόʔ ηΩϡϦςΟઓུͷ؂ࢹʹऔΓ૊Ή΋ͷ • ૊৫ͷϛογϣϯͱར֐ؔ܎ऀͷظ଴ʹরΒ͠ ͯɺଞͷ5ͭͷػೳͷ੒ՌΛୡ੒͠ɺ༏ઌॱҐΛ ͚ͭΔͨΊʹ૊৫͕ԿΛ͢΂͖͔ࣔͨ͢Ίͷ੒ ՌΛఏڙ͢Δ
22. What is governance? • NIST CSF 2.0ʹ͓͚ΔʮGovernʢ౷࣏ʣʯͷΧςΰϦ • ૊৫తจ຺ • ϦεΫϚωδϝϯτઓུ • ໾ׂ / ੹೚ / ݖݶ • ϙϦγʔ • ؂ಜ • αΠόʔηΩϡϦςΟαϓϥΠνΣʔϯϦεΫϚωδϝϯτ
23. ͳΔ΄ͲΘ͔ΒΜʁ
24. ཁ͸ର৅ͷࣄฑʹରͯ͠ ઓུʹج͍ͮͨϙϦγʔΛ࡞Γ ϙϦγʔΛ਱ߦ͢ΔͨΊͷ૊৫ΛఆΊ ੹೚/໾ׂ/ݖݶΛׂΓৼΓ ؂ࢹ͢Δ͜ͱ
25. ੜ੒AIͷΨόφϯεͷͨΊʹ ԿΛ͠ͳ͍ͱ͍͚ͳ͍ͷʁ
26. Governance Guidelines for AI Principles of Practice. • ؀ڥɾϦεΫ෼ੳ • AIΨόφϯεΰʔϧઃఆ • AIϚωδϝϯτγεςϜઃܭ • ܧଓతͳධՁ • ΞδϟΠϧΨόφϯεͷ࣮ફ "*
    ݪଇ࣮ફͷͨΊͷ
    ΨόφϯεɾΨΠυϥΠϯ
    7FS
    
    IUUQTXXXNFUJHPKQTIJOHJLBJNPOP@JOGP@TFSWJDFBJ@TIBLBJ@KJTTPQEG@QEG
27. Governance Guidelines for AI Principles of Practice. • ؀ڥɾϦεΫ෼ੳ • اۀ΍ࣄۀ෦୯Ґͷํ਑ΛܾΊΔʹ͋ͨͬͯ͸ɺAI γεςϜ͕΋ ͨΒ͠͏Δਖ਼ෛͷΠϯύΫτɺAI γεςϜͷ։ൃ΍ӡ༻ʹؔ͢Δ ࣾձతड༰ɺͦͯࣗࣾ͠ͷࣄۀൣғ౳ʹরΒͯ͠ෛͷΠϯύΫτ ͕ܰඍͰ͸ͳ͍ͱ൑அͨ͠৔߹ʹ͸ɺࣗࣾͷ AI शख़౓ʢAI γε ςϜͷ։ൃɾӡ༻࣌ʹٻΊΒΕΔ४උ͕ͲΕ͚ͩͰ͖͍ͯΔͷ ͔ʣΛߟྀ͢΂͖Ͱ͋Δɻ
28. Governance Guidelines for AI Principles of Practice. • AIΨόφϯεΰʔϧઃఆ • ʢͨͱ͑͹ AI ϙϦγʔʣΛઃఆ͢Δ͔൱͔ʹ͍ͭͯݕ౼͢΂͖ Ͱ͋ΓɺજࡏతͳෛͷΠϯύΫτ͕ܰඍͰ͋Δ͜ͱΛཧ༝ʹ AI ΨόφϯεɾΰʔϧΛઃఆ͠ͳ͍৔߹ʹ͸ɺͦͷཧ༝౳Λεςʔ Ϋϗϧμʔʹઆ໌Ͱ͖ΔΑ͏ʹ͓ͯ͘͠΂͖Ͱ͋Δɻ
29. Governance Guidelines for AI Principles of Practice. • γεςϜσβΠϯʢAIϚωδϝϯτγεςϜઃܭʣ • ΰʔϧΛୡ੒͢ΔͨΊͷઃܭ • AI Ψόφϯεɾΰʔϧ͔Βͷဃ཭ͷධՁͱဃ཭΁ͷରԠ • AI ϚωδϝϯτγεςϜΛ୲͏ਓࡐͷϦςϥγʔ޲্ • ద੾ͳ৘ใڞ༗ͳͲͷࣄۀऀؒɾ෦໳ؒڠྗʹΑΔ AI Ϛωδϝϯτͷ ڧԽͱΠϯγσϯτͷ༧๷΍ૣظରԠΛ௨ͨ͡ΠϯγσϯτʹؔΘΔ AI γεςϜར༻ऀͷෛ୲ܰݮ
30. Governance Guidelines for AI Principles of Practice. • ܧଓతͳධՁ • AI ϚωδϝϯτγεςϜٴͼݸʑͷ AI γεςϜͷӡ༻ঢ়گʹͭ ͍ͯઆ໌Մೳͳঢ়ଶΛ֬อ͢Δ (಺෦؂ࠪ, ࣗݾධՁ, ֎෦؂ࠪ) • εςʔΫϗϧμʔͱͷҰ૚ԁ׈ͳίϛϡχέʔγϣϯͷͨΊʹɺ ͜ΕΒͷ৘ใΛίʔϙϨʔτΨόφϯεɾίʔυͷඇࡒ຿৘ใʹ Ґஔ͚ͮɺੵۃతʹ։ࣔ͢Δ͜ͱΛݕ౼͢΂͖
31. Governance Guidelines for AI Principles of Practice. • ΞδϟΠϧΨόφϯεͷ࣮ફ • εςʔΫϗϧμʔͷؔ༩ͷ ԼͰɺAIΨόφϯεͷࡏΓ ํͷݕ౼Λܧଓ͠ɺඞཁʹ Ԡͯ͡վగΛߦ͏͜ͱ͕๬ ·͍͠ɻ "*
    ݪଇ࣮ફͷͨΊͷ
    ΨόφϯεɾΨΠυϥΠϯ
    7FS
    
    IUUQTXXXNFUJHPKQTIJOHJLBJNPOP@JOGP@TFSWJDFBJ@TIBLBJ@KJTTPQEG@QEG
32. Μʔɺࠓͻͱͭந৅తͰ Ͱ͖Δؾ͕͠ͳ͍
33. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτɾڴҖͷϞσϦϯά • ૊৫ମ੍ͷཱ֬ʢ੹೚ऀɺઆ໌੹೚ऀɺ૬ஊ໾ɺใࠂઌʣ • AIϦεΫɺϦεΫධՁɺ૊৫಺ͷΨόφϯε੹೚ͷจॻԽͱ୲౰ऀͷׂΓ౰ͯ • σʔλ؅ཧϙϦγʔɺAIϙϦγʔʢળߦج४ɺσʔλอޢɺιϑτ΢ΣΞ࢖༻ʣ • ࢖ͬͯྑ͍ੜ੒AIπʔϧҰཡ • LLMੜ੒ϞσϧͰ࢖͏σʔλͷιʔεͱ؅ཧͷจॻԽ --.
    "*
    αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ
    ʙ
    ࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ
    ʙ
    IUUQTPXBTQPSHXXXQSPKFDUUPQGPSMBSHFMBOHVBHFNPEFMBQQMJDBUJPOTMMNUPQHPWFSOBODFEPD--.@"*@4FDVSJUZ@BOE@(PWFSOBODF@$IFDLMJTUW@@+1QEG
34. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM
35. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM • ఢରऀʹوॏͳ৘ใΛఏڙ͢ΔՄೳੑ ͷ͋Δ͢΂ͯͷσʔλɺࢿ࢈ɺݸਓ৘ใ ʹ͍ͭͯɺ୯ಠɾू߹໰Θͣɺ૊৫తͳ ཧղΛਂΊΔ
36. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM • ෺ཧతɺαΠόʔɺ·ͨ͸ӡ༻্ͷϦ εΫΛ΋ͨΒ͢Մೳੑͷ͋ΔڴҖߦҝ ऀ͕࢖༻͢Δઓज़Λཧղ͢Δ
37. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM • ಛఆͨ͠ػີ৘ใ΁ͷΞΫηεΛఢʹ ڐ͢Մೳੑͷ͋Δɺ෺ཧత͓ΑͼαΠ όʔηΩϡϦςΟखॱͷજࡏతͳ੬ऑ ੑΛಛఆ͢Δ
38. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM • ڴҖͱ੬ऑੑΛߟྀͯ͠ɺಛఆͨ͠ػ ີ৘ใʹΞΫηεͰ͖ͨ৔߹ɺબڍΠ ϯϑϥ΍ϓϩηεͷηΩϡϦςΟʹର ͢ΔڴҖߦҝऀͷߦಈͷՄೳੑͱॏେ ੑΛධՁ͢Δ
39. Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ػີ৘ใͷಛఆ • ڴҖΛཧղ͢Δ • ੬ऑੑͷಛఆ • ϦεΫͷධՁ • ରࡦͷ࣮ࢪ 014&$ ΦϖϨʔγϣφϧηΩϡϦςΟ ݪଇͷ࣮ࢪ
    IUUQTJOUFSOFUXBUDIJNQSFTTDPKQEPDTDPMVNOTFDVSJUZIUNM • ಛఆͨ͠༏ઌϦεΫͷআڈ΍௿ݮ͢Δ ରࡦΛબ୒ͯ͠ɺ࣮ࢪ͢Δ
40. Elements in putting together an AI governance structure. • ڴҖͷϞσϦϯά • ੜ੒AIΛ༻͍ͨେن໛Ͱߴ౓ͳʮͳΓ͢·͠߈ܸʯ( ϑΟογϯάɺεϓʔϑΟϯά) • LLMιϦϡʔγϣϯ΁ͷ༗֐·ͨ͸ѱҙͷ͋Δೖྗ΍ΫΤϦ • LLMͷ͢΂ͯͷ৴པڥքͰɺϞσϧ΍಺෦ͷγεςϜͱσʔλϕʔε΁ͷ֎෦͔Β ͷෆਖ਼ΞΫηε • ڐՄ͞ΕͨϢʔβʔʹΑΔޡ༻ ( ೖྗɾग़ྗͷίϯςϯπϑΟϧλϦϯά ) --.
    "*
    αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ
    ʙ
    ࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ
    ʙ
    IUUQTPXBTQPSHXXXQSPKFDUUPQGPSMBSHFMBOHVBHFNPEFMBQQMJDBUJPOTMMNUPQHPWFSOBODFEPD--.@"*@4FDVSJUZ@BOE@(PWFSOBODF@$IFDLMJTUW@@+1QEG
41. [re]: Risks posed by AI • όΠΞεͷ͋Δ݁Ռ΍ࠩผతͳ݁Ռͷग़ྗ • ϑΟϧλʔόϒϧɺΤίʔνΣϯόʔݱ৅ • ଟ༷ੑͷ૕ࣦ • ෆద੾ͳݸਓ৘ใͷऔѻ͍ • ੜ໋ɾ਎ମɾࡒ࢈ͷ৵֐ • ϒϥοΫϘοΫεԽɺ൑அʹؔ͢Δઆ໌ཁٻ • ΤωϧΪʔ࢖༻ྔٴͼ؀ڥͷෛՙ • ػີ৘ใͷྲྀग़ • ѱ༻ • ϋϧγωʔγϣϯ • ِ৘ใɺޡ৘ใΛӏವΈʹ͢Δ͜ͱ • ஶ࡞ݖͱͷؔ܎ • ࢿ֨౳ͱͷؔ܎ • όΠΞεͷ࠶ੜ੒ "*ࣄۀऀΨΠυϥΠϯʢୈ൛ʣ
    IUUQTXXXNFUJHPKQQSFTTIUNM
42. Elements in putting together an AI governance structure. • ૊৫ମ੍ͷཱ֬ʢ੹೚ऀɺઆ໌੹೚ऀɺ૬ஊ໾ɺใࠂઌʣ • ITɺηΩϡϦςΟɺ๏຿֤෦໳ͱͷ࿈ܞ͕ඞਢ • LLMϞσϧ΍LLMαʔϏεΛఏڙ͢Δ৔߹ɺطଘͷར༻ن໿ ͷݟ௚͠ͱߋ৽ɺEULAܖ໿ॻΛ࡞ΓࠐΉ • ηϯγςΟϒ෼໺ʹ͓͍ͯ͸ɺ։ൃɺӡ༻ɺ๏຿෦໳ͷ੹೚ऀ͔ ΒͳΔԣஅతͳձٞ --.
    "*
    αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ
    ʙ
    ࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ
    ʙ
    IUUQTPXBTQPSHXXXQSPKFDUUPQGPSMBSHFMBOHVBHFNPEFMBQQMJDBUJPOTMMNUPQHPWFSOBODFEPD--.@"*@4FDVSJUZ@BOE@(PWFSOBODF@$IFDLMJTUW@@+1QEG
43. [re]: Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ૊৫ମ੍ͷཱ֬ʢ੹೚ऀɺઆ໌੹೚ऀɺ૬ஊ໾ɺใࠂઌʣ • AIϦεΫɺϦεΫධՁɺ૊৫಺ͷΨόφϯε੹೚ͷจॻԽͱ୲౰ऀͷׂΓ౰ͯ • σʔλ؅ཧϙϦγʔɺAIϙϦγʔʢળߦج४ɺσʔλอޢɺιϑτ΢ΣΞ࢖༻ʣ • ࢖ͬͯྑ͍ੜ੒AIπʔϧҰཡ • LLMੜ੒ϞσϧͰ࢖͏σʔλͷιʔεͱ؅ཧͷจॻԽ --.
    "*
    αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ
    ʙ
    ࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ
    ʙ
    IUUQTPXBTQPSHXXXQSPKFDUUPQGPSMBSHFMBOHVBHFNPEFMBQQMJDBUJPOTMMNUPQHPWFSOBODFEPD--.@"*@4FDVSJUZ@BOE@(PWFSOBODF@$IFDLMJTUW@@+1QEG
44. ݁ہɺϧʔϧΛ࡞ͬͯ ਓྗͰӡ༻ͬͯ͜ͱʁ
45. ݱࡏ஍ͱͯ͠͸ͦ͏
46. ͜Ε͔Β
47. [re]: Elements in putting together an AI governance structure. • ϦεΫΞηεϝϯτ • ૊৫ମ੍ͷཱ֬ʢ੹೚ऀɺઆ໌੹೚ऀɺ૬ஊ໾ɺใࠂઌʣ • AIϦεΫɺϦεΫධՁɺ૊৫಺ͷΨόφϯε੹೚ͷจॻԽͱ୲౰ऀͷׂΓ౰ͯ • σʔλ؅ཧϙϦγʔɺAIϙϦγʔʢળߦج४ɺσʔλอޢɺιϑτ΢ΣΞ࢖༻ʣ • ࢖ͬͯྑ͍ੜ੒AIπʔϧҰཡ • LLMੜ੒ϞσϧͰ࢖͏σʔλͷιʔεͱ؅ཧͷจॻԽ --.
    "*
    αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ
    ʙ
    ࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ
    ʙ
    IUUQTPXBTQPSHXXXQSPKFDUUPQGPSMBSHFMBOHVBHFNPEFMBQQMJDBUJPOTMMNUPQHPWFSOBODFEPD--.@"*@4FDVSJUZ@BOE@(PWFSOBODF@$IFDLMJTUW@@+1QEG
48. Future Governance of Generative AI • ϦεΫΞηεϝϯτ • ੜ੒AIͱೖग़ྗͷؔ܎ੑ ΛਤࣜԽͨ͠ϑϨʔϜ ϫʔΫʹϢʔεέʔεΛ ౰ͯ͸ΊΔ͜ͱͰɺϦε ΫΛΠϝʔδ͠қ͘͢Δ ੜ੒"*αʔϏεར༻ʹؔ͢Δ஫ҙϙΠϯτݕग़ϑϨʔϜϫʔΫΛߟ͑ͯΈͨ
    IUUQTXXXGOJGOJOFUBJGSBNFXPSL
49. • CoE ( Center of Excellence ) ͷઃஔ • φϨοδͷऩूͱ੔ཧ • Ψόφϯεͱϓϩηεͷ੔උ • τϨʔχϯάͱٕज़ࢧԉ • ૊৫શମΛ၆ᛌͨ͠ϑΟʔυόοΫ • ϓϥΠόγʔͱσʔλΨόφϯεͷૠ͠Ͳ͜Ζ Future Governance of Generative AI
50. Future Governance of Generative AI • ࢖ͬͯྑ͍ੜ੒AIπʔϧҰཡ • ໺ྑੜ੒AIΛձࣾͷੜ੒ AIʹ༠ಋ • ར༻ΛೝՄ͞Εͨੜ੒AI ΁ͷೖྗσʔλͷαχλ Πζ
51. Future Governance of Generative AI • LLMੜ੒AIϞσϧͰ࢖͏σʔλͷιʔεͱ؅ཧ • AWS ੜ੒ AI ϕετϓϥΫςΟεϑϨʔϜϫʔΫ v2 • https://docs.aws.amazon.com/ja_jp/audit-manager/latest/ userguide/aws-generative-ai-best-practices.html • ੜ੒ AI ͷͨΊͷωοτϫʔΫڥքͰͷηΩϡϦςΟอޢ • https://aws.amazon.com/jp/blogs/news/network-perimeter- security-protections-for-generative-ai/
52. Conclusion • ٕज़΁ͷཧղ͸Ψόφϯεʹد༩͢Δ͠ɺٕज़΁ͷཧղͳͯ͘͠ Ψόφϯε͸ୡ੒͠ಘͳ͍ • ҰํͰɺݸʑͷٕज़͸ɺΨόφϯεͳͯ͘͠े෼ʹػೳ͠ͳ͍ • ·ͣ͸৘ใηΩϡϦςΟͷجຊΛ཈͑Δͱ͜Ζ͔Β࢝ΊΔ • Ϗδωεಛੑͱ૊৫ΛऔΓר͘؀ڥ΍๏ن੍ͷཧղͱ੔ཧ • ػີ৘ใͷࣝผͱॴࡏͷ೺ѲɺϦεΫΞηεϝϯτ
53. Thank you !
54. Appendix • LLM AI αΠόʔηΩϡϦςΟͱΨόφϯεͷνΣοΫϦετ ʙࣦഊ͠ͳ͍େن໛ݴޠϞσϧಋೖͷͨΊʹ ʙ • https://owasp.org/www-project-top-10-for-large-language-model-applications/llm-top-10-governance-doc/ LLM_AI_Security_and_Governance_Checklist-v1_1_JP.pdf • OPSEC(ΦϖϨʔγϣφϧηΩϡϦςΟ)ݪଇͷ࣮ࢪ • https://internet.watch.impress.co.jp/docs/column/security/1614011.html • AI ݪଇ࣮ફͷͨΊͷ ΨόφϯεɾΨΠυϥΠϯ Ver. 1.1 • https://www.meti.go.jp/shingikai/mono_info_service/ai_shakai_jisso/pdf/20220128_1.pdf • AIࣄۀऀΨΠυϥΠϯʢୈ1.0൛ʣ • https://www.meti.go.jp/press/2024/04/20240419004/20240419004.html
55. Appendix • ੜ੒ AI αʔϏεͷར༻ʹؔ͢Δ஫ҙשى౳ʹ͍ͭͯ • https://www.ppc.go.jp/ fi les/pdf/230602_kouhou_houdou.pdf • OWASP Top 10 for LLM Λ׆༻ͨ͠ੜ੒ AI ΞϓϦέʔγϣϯͷଟ૚๷ޚηΩϡϦςΟઃܭ • https://aws.amazon.com/jp/blogs/news/architect-defense-in-depth-security-for-generative-ai-applications-using-the- owasp-top-10-for-llms/ • AIࣄۀऀΨΠυϥΠϯʢୈ1.0൛ʣ • https://www.meti.go.jp/press/2024/04/20240419004/20240419004.html • The NIST Cybersecurity Framework (CSF) 2.0 • https://nvlpubs.nist.gov/nistpubs/CSWP/NIST.CSWP.29.pdf
56. Appendix • ϓϩϑΝΠϦϯάʹؔ͢Δ࠷ऴఏݴ ࣗओతऔ૊Έʹؔ͢ΔνΣοΫϦετ - ύʔιφϧσʔλ+αݚڀձ • https://wp.shojihomu.co.jp/wp-content/uploads/2022/04/ef8280a7d908b3686f23842831dfa659.pdf • DX࣌୅ʹ͓͚ΔاۀͷϓϥΠόγʔΨόφϯεΨΠυϒοΫver1.1 • https://www.meti.go.jp/policy/it_policy/privacy/guidebook11gaiyo.pdf • ίʔϙϨʔτΨόφϯεɾίʔυ • https://www.jpx.co.jp/equities/listing/cg/tvdivq0000008jdy-att/nlsgeu000005lnul.pdf • ੜ੒AIαʔϏεར༻ʹؔ͢Δ஫ҙϙΠϯτݕग़ϑϨʔϜϫʔΫΛߟ͑ͯΈͨ • https://www.fnifni.net/ai-framework/