/AI3h ago

Nathan Lambert, AI2 post-training lead, argues Anthropic's inconsistent safety policies for Claude Fable 5 weaken collective AI safety

Ravid Shwartz Ziv accused Anthropic of using safety to block competition.

18277253715.3K
Original post
Nathan Lambert@natolambert#64inAI

Why I think Anthropic's uneven safety policies with the release of Claude Fable 5 undermine the broader AI community's cohesion and accelerate us to more uncertainty and risk in AI's near-term evolution. https://www.interconnects.ai/p/claude-fable-5-and-new-ai-safety

4:00 PM · Jun 9, 2026 · 11.1K Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS2.3K
Nathan Lambert@natolambert

(i've been pretty sick the last few days so if something is off plz be charitable in the rapid speed blog assembly line)

Nathan Lambert@natolambert

Why I think Anthropic's uneven safety policies with the release of Claude Fable 5 undermine the broader AI community's cohesion and accelerate us to more uncertainty and risk in AI's near-term evolution. https://www.interconnects.ai/p/claude-fable-5-and-new-ai-safety

3hViews 2.3KLikes 9Bookmarks 0
BOOKMARKS3LIKES36REPLIES2

Great article by Nathan. Antropic plays a double-standard game here (and in many other cases). They talk a lot about safety (which is super important), but what they actually do is prevent competition (which is legit, but should be transparent about it). They are wrapping the two together because they want to look good

Nathan Lambert@natolambert

Why I think Anthropic's uneven safety policies with the release of Claude Fable 5 undermine the broader AI community's cohesion and accelerate us to more uncertainty and risk in AI's near-term evolution. https://www.interconnects.ai/p/claude-fable-5-and-new-ai-safety

2hViews 2.3KLikes 36Bookmarks 3

Great article by Nathan. Antropic plays a double-standard game here (and in many other cases). A double staney talked a lot about safety (which is super important), but what they actually do is prevent competition (which is legit, but should be transparent about it). They are wrapping the two together because they want to look good

Nathan Lambert@natolambert

Why I think Anthropic's uneven safety policies with the release of Claude Fable 5 undermine the broader AI community's cohesion and accelerate us to more uncertainty and risk in AI's near-term evolution. https://www.interconnects.ai/p/claude-fable-5-and-new-ai-safety

2hViews 115Likes 2Bookmarks 1
rallio@w_bsun

I understand generic griping, but you are complaining about a company that has consistently been against open source, consistently had very restrictive terms of service, and consistently argued against diffusion of the cutting edge technology. You can dislike it all you want, but this is par for the course from Anthropic since the earliest days.

3hViews 119
Nathan Lambert@natolambert

@w_bsun Mostly I've given them a pass because it was mostly being annoying before and not taking action.

3hViews 85Likes 4
POM@peterom

@natolambert Fable agrees:

3hViews 76Likes 1
Kazba@Kazba22

@natolambert the tension between moving fast and holding the line was always going to crack something eventually

3hViews 20Likes 1
haro@harobuilds

@natolambert the irony is that every major lab ships safety comms that read like they were written for a different product than the one in the release notes. fable 5 isn't unique in that gap, it's just the most visible example right now

3hViews 1Likes 1