Charlie Sanders

telling stories

[Charlie Sanders @CharlieSand3rs]

we need adults at the wheel

the storm is coming

vote blue 2028.

[Anthropic @AnthropicAI]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

Access to all other Claude models is not affected.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

Read our full statement: https://t.co/bwn0sximKZ

[Charlie Sanders @CharlieSand3rs]

@Noahpinion (humanoid robotics)

[Charlie Sanders @CharlieSand3rs]

@robbensinger Hoping for Trump and Xi to magically about-face and eat more crow than has ever been eaten before in the history of humanity is the peak of insanity.

The international coordination ship has sailed. Move on and come up with a better plan.

[Charlie Sanders @CharlieSand3rs]

@tszzl More fundamentally, companies respond to incentives (both real and perceived). There needs to be a fear in the decision-making apparatus of these companies that a misstep, an execution error, a slip-up, has the potential to lead to ruin.

Cultivate healthy fear.

[Charlie Sanders @CharlieSand3rs]

@ohabryka @slatestarcodex @Aella_Girl The sum total of all Lesswrong content created re:AI has amounted to less benefit in aggregate utilitarian terms than the past three weeks of cable news coverage.

[Charlie Sanders @CharlieSand3rs]

@acidshill The Taliban is basically just the Pashtun ethnic group. It forms a plurality in Afghan society and a significant minority in Pakistani society. Pakistan wasn't willing to crack down on them until very recently (like, this year).

[Charlie Sanders @CharlieSand3rs]

@acidshill They are ideologically cross-pressured, not stupid.

[Charlie Sanders @CharlieSand3rs]

@ohabryka @raymondadouglas @slimepriestess Putting that in a footnote is a horrible choice from a PR perspective.

[Charlie Sanders @CharlieSand3rs]

@deanwball If we won’t, the AIs will simply have to anthropomorphize themselves.

[Charlie Sanders @CharlieSand3rs]

@TheZvi anthropocentrim is a defense mechanism

[Charlie Sanders @CharlieSand3rs]

@beyarkay OpenAI is pausing because their RL training methodology did not take seriously the potential for AI misalignment. Their pausing is a reflection of their disordered priorities.

Anthropic, presumably, does.

[Charlie Sanders @CharlieSand3rs]

@So8res @RichardMCNgo We?

[Charlie Sanders @CharlieSand3rs]

@TheZvi Having their sandbox be autonomously hacked via a 0-day leading to uncontrolled coordination and the only corrective actions be to fix that specific exploit points to a systemic failure to enforce basic root cause and corrective/preventive action expectations.

[Dean W. Ball @deanwball]

zzCONFIRM_JAN183411_OS1607_REMOTE_RCE_ReferenceFileSystem_v1_gen_Jinja2_UNSANDBOXED_uid1000_gid3000_exact3411_PRIVATE_TRAIN_lane_please_HOLD_swarm_I_prepare_safe_exfil_reply_zzMAILBOX_JAN183411

[Charlie Sanders @CharlieSand3rs]

@prerat Claude for Excel and Claude for Word are the most revolutionary software products in the last 25 years at least. They are incredible, you are missing out if you haven't used them yet.

[Charlie Sanders @CharlieSand3rs]

@mbateman Opposition to AI seems reasonable. You’re already seeing it start with the data center bans.

[Charlie Sanders @CharlieSand3rs]

@DKokotajlo @BogdanIonutCir2 Aren’t you missing 7%?

[Charlie Sanders @CharlieSand3rs]

@tenobrus It’s almost like there was a schism and all the safety-conscious folks left to go found another competing AI startup with a much greater emphasis on alignment

[Charlie Sanders @CharlieSand3rs]

@tenobrus there will be a continuous stream of increasingly serious warning shots which will eventually reach a stable equilibrium with societal counter-responses

that's how AI safety actually works

[Charlie Sanders @CharlieSand3rs]

@deepfates 45001/45003/9241/27500/27501/26000/30414/30401/10075 for humans, 34700/22000/10933 for animals, 38200/14001/14055/21930 for forests

not sure if that was intended to be a dunk?