Setelah larangan selama dua minggu, pemerintah AS mengizinkan Anthropic mengirimkan kembali model AI terkuatnya secara global.
Fable 5 kembali mendunia mulai hari ini melalui Platform Claude, Claude.ai, Claude Code, dan Claude Cowork. Paket Pro, Max, Team, dan Enterprise tertentu menyertakan model ini hingga 7 Juli dengan batas penggunaan mingguan hingga 50 persen. Setelah itu, akan ditagih melalui kredit penggunaan. Akses di AWS, Google Cloud, dan microsoft Foundry akan dipulihkan "secepat mungkin".
Mitos 5, versi yang tidak terlalu dibatasi dari model dasar yang sama, tetap terbatas pada sekelompok organisasi AS yang mendapat persetujuan pemerintah pada tanggal 26 Juni. Anthropic mengatakan pihaknya masih bekerja sama dengan pemerintah untuk memperluas akses ke lebih banyak mitra dalam program yang disebut Glasswing. Apakah UE akan bergabung masih belum jelas.Ad
Anthropic menegaskan bahwa larangan tersebut berasal dari temuan keamanan oleh para peneliti Amazon. Mereka telah menemukan cara untuk melewati pagar pengaman Fable 5. Model tersebut kemudian mengidentifikasi beberapa kerentanan perangkat lunak dan, dalam satu kasus, menghasilkan kode yang menunjukkan cara mengeksploitasi salah satunya.
Pemerintah AS dan Anthropic menghabiskan waktu dua minggu untuk menyelidiki kerentanan tersebut. Banyak model yang kurang mampu dapat menemukan kelemahan yang sama seperti yang ditemukan dalam laporan Fable 5, termasuk Claude Opus 4.8, GPT-5.5, dan Kimi K2.7. Untuk demo eksploitasi spesifik, setiap model yang diuji menghasilkan hasil yang sama, bahkan model yang jauh lebih kecil seperti Claude Haiku 4.5.
Anthropic calls it an edge case that only involved routine defensive cybersecurity work. In response, the company trained an improved safety classifier that blocks the technique from the Amazon report in more than 99 percent of cases. When a request gets blocked, users see a notification, and the request gets routed to the older Opus 4.8 model.Ad
The new classifier comes with a tradeoff, though. It flags harmless requests more often during everyday coding and debugging. Users had already complained the model wastoo restrictive during the first Fable release.
No universal jailbreak was found at the time of release. But the company admits it's "probably impossible to make any ai model fully robust (that is, impervious) to jailbreaks." That was well known before Fable 5 shipped.Ad
The AI industry needs a shared standard for rating jailbreaks and triggering countermeasures, Anthropic argues. The company says it's building such a framework with Amazon, Microsoft, Google, and other Glasswing partners. Anthropic is also standing up a team for 24/7 monitoring of jailbreak submission channels and launched a newHackerOne programwhere security researchers can report potential cyber jailbreaks for Fable 5.Ad
Anthropic is expanding its work with the US government, building on their joint efforts tied to theexecutive order.
The company is making several commitments. Government partners will get pre-release access to models that advance capabilities in security-sensitive areas. Discovered jailbreaks or abuse patterns will be shared quickly. Anthropic will put up dedicated resources and significant compute for joint research. And the company will help build a shared industry standard for frontier model providers.
Anthropic wants all of this written into "strong regulation" and applied equally to every frontier model developer. "Government involvement in ai releases requires a durable, transparent process that gives cyber defenders and others the certainty they need about access to powerful models," the company writes.
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.