After spooking Hollywood, ByteDance will tweak safeguards on new AI model
After spooking Hollywood, ByteDance will tweak safeguards on new AI model
Its easier to apologise than ask permission.
buff.ly
February 16, 2026 at 11:40 AM
After spooking Hollywood, ByteDance will tweak safeguards on new AI model
At the risk of making this boring, the entire point of administrative law is to prevent agencies like the EPA from just making shit up like this. Unfortunately, the safeguards for accomplishing this were primarily norms-based and at best reactive.
New in PN: "Just making s**t up" — @volts.wtf sounds off on EPA
“How the f**k are they gonna do it? If you can get the Court to rule carbon is not a public health danger, it'll mark the point at which they've abandoned all pretense, bc that’s just a flat out lie” www.publicnotice.co/p/david-robe...
“How the f**k are they gonna do it? If you can get the Court to rule carbon is not a public health danger, it'll mark the point at which they've abandoned all pretense, bc that’s just a flat out lie” www.publicnotice.co/p/david-robe...
February 16, 2026 at 3:42 PM
At the risk of making this boring, the entire point of administrative law is to prevent agencies like the EPA from just making shit up like this. Unfortunately, the safeguards for accomplishing this were primarily norms-based and at best reactive.
Shoutout to this mod I've been trying. The good news it removes all the automated safeguards that keep your ranged units from shooting. The bad news it removes all the automated safeguards that keep your units from shooting
February 17, 2026 at 3:39 PM
Shoutout to this mod I've been trying. The good news it removes all the automated safeguards that keep your ranged units from shooting. The bad news it removes all the automated safeguards that keep your units from shooting
President Trump’s so-called Department of War is threatening to cut ties with Anthropic because it won’t help the Pentagon conduct mass surveillance on Americans.
It warned Anthropic it will “pay a price” if it continues to demand its products have safeguards. trib.al/rDDI5h3
It warned Anthropic it will “pay a price” if it continues to demand its products have safeguards. trib.al/rDDI5h3
February 17, 2026 at 2:41 PM
President Trump’s so-called Department of War is threatening to cut ties with Anthropic because it won’t help the Pentagon conduct mass surveillance on Americans.
It warned Anthropic it will “pay a price” if it continues to demand its products have safeguards. trib.al/rDDI5h3
It warned Anthropic it will “pay a price” if it continues to demand its products have safeguards. trib.al/rDDI5h3
New "boundary point jailbreaking" method against LLM safeguards (with prior disclosure to multiple labs) by using noised versions of harmful queries to turn sparse feedback from failed attacks into dense feedback. 🧵
www.aisi.gov.uk/blog/boundar...
www.aisi.gov.uk/blog/boundar...
February 17, 2026 at 8:55 PM
New "boundary point jailbreaking" method against LLM safeguards (with prior disclosure to multiple labs) by using noised versions of harmful queries to turn sparse feedback from failed attacks into dense feedback. 🧵
www.aisi.gov.uk/blog/boundar...
www.aisi.gov.uk/blog/boundar...
Exactly one year ago, Google removed "weapons and surveillance" from non-permitted uses in its AI principles.
It is now clear why:
"OpenAI, Google and xAI, all... have agreed to remove their safeguards for use in the military's unclassified systems."
www.axios.com/2026/02/16/a...
It is now clear why:
"OpenAI, Google and xAI, all... have agreed to remove their safeguards for use in the military's unclassified systems."
www.axios.com/2026/02/16/a...
Exclusive: Pentagon warns Anthropic will "pay a price" as feud escalates
The Pentagon may label Anthropic a "supply chain risk," forcing all its vendors to sever ties.
www.axios.com
February 18, 2026 at 2:04 PM
Exactly one year ago, Google removed "weapons and surveillance" from non-permitted uses in its AI principles.
It is now clear why:
"OpenAI, Google and xAI, all... have agreed to remove their safeguards for use in the military's unclassified systems."
www.axios.com/2026/02/16/a...
It is now clear why:
"OpenAI, Google and xAI, all... have agreed to remove their safeguards for use in the military's unclassified systems."
www.axios.com/2026/02/16/a...
Pentagon threatens to cut off Anthropic in AI safeguards dispute, Axios reports reut.rs/4cuv9Y8
Pentagon threatens to cut off Anthropic in AI safeguards dispute, Axios reports
The Pentagon is considering severing its relationship with Anthropic over the artificial intelligence firm's insistence on maintaining some limitations on how the U.S. military uses its models, Axios reported on Saturday, citing an administration official.
reut.rs
February 15, 2026 at 2:30 AM
Pentagon threatens to cut off Anthropic in AI safeguards dispute, Axios reports reut.rs/4cuv9Y8
A major issue with “the left refuses to engage with AI” is that, at heart, the industry wants complete deregulation and zero safeguards, and there’s no meaningful way for the left to engage with that. Engaging with the technology itself is a secondary or maybe even tertiary issue.
February 17, 2026 at 7:14 PM
A major issue with “the left refuses to engage with AI” is that, at heart, the industry wants complete deregulation and zero safeguards, and there’s no meaningful way for the left to engage with that. Engaging with the technology itself is a secondary or maybe even tertiary issue.
Three years after the feature rolled out, Apple is automatically enabling Stolen Device Protection in iOS 26.4, expanding safeguards against an ever-expanding epidemic of iPhone theft.
iOS 26.4 has iPhone Stolen Device Protection on by default
Three years after the feature rolled out, Apple is automatically enabling Stolen Device Protection in iOS 26.4, expanding safeguards against an ever-expanding epidemic of iPhone theft.
appleinsider.com
February 16, 2026 at 8:12 PM
Three years after the feature rolled out, Apple is automatically enabling Stolen Device Protection in iOS 26.4, expanding safeguards against an ever-expanding epidemic of iPhone theft.
More than 100 researchers call for safeguards on high-risk biological datasets to prevent AI misuse, which could create deadly pathogens.
Experts warn open data may help AI design dangerous pathogens
More than 100 researchers call for safeguards on high-risk biological datasets to prevent AI misuse, which could create deadly pathogens.
l.euronews.com
February 18, 2026 at 11:54 PM
More than 100 researchers call for safeguards on high-risk biological datasets to prevent AI misuse, which could create deadly pathogens.
Rep. Amy Walen (D-48th, Kirkland) joins the Republican caucus in voting against a bill regulating AI chatbots. The bill requires chatbot designers to add safeguards when it comes to users who express suicidal ideation and self-harm.
app.leg.wa.gov/billsummary/...
app.leg.wa.gov/billsummary/...
HB 2225 Washington State Legislature
app.leg.wa.gov
February 17, 2026 at 9:49 PM
Rep. Amy Walen (D-48th, Kirkland) joins the Republican caucus in voting against a bill regulating AI chatbots. The bill requires chatbot designers to add safeguards when it comes to users who express suicidal ideation and self-harm.
app.leg.wa.gov/billsummary/...
app.leg.wa.gov/billsummary/...
“It’s horrifying that political attacks on environmental safeguards could erase the existence of an animal as ancient and magnificent as the hellbender,” said Tierra Curry, our #EndangeredSpecies co-director.
February 17, 2026 at 7:03 PM
“It’s horrifying that political attacks on environmental safeguards could erase the existence of an animal as ancient and magnificent as the hellbender,” said Tierra Curry, our #EndangeredSpecies co-director.
I will never not believe that OpenAI just pushing ChatGPT in the world without safeguards or regard to customers will go down as one of the most consequential decisions in technology history. ML and AI was being worked on and improved, and others were being responsible....and then boom.
February 18, 2026 at 3:22 PM
I will never not believe that OpenAI just pushing ChatGPT in the world without safeguards or regard to customers will go down as one of the most consequential decisions in technology history. ML and AI was being worked on and improved, and others were being responsible....and then boom.
Alongside our partners, we’re suing the EPA for illegally abandoning its duty to protect us from climate pollution. Its unscientific repeal of vital safeguards defies the law, science, and public interest. We’ll see them in court, and we will win.
Public health and green groups sue EPA over repeal of rule supporting climate protections
A coalition of health and environmental groups is suing the Environmental Protection Agency, challenging the rescinding of a scientific finding that has been the central basis for U.S. action to regul...
apnews.com
February 19, 2026 at 12:06 AM
Alongside our partners, we’re suing the EPA for illegally abandoning its duty to protect us from climate pollution. Its unscientific repeal of vital safeguards defies the law, science, and public interest. We’ll see them in court, and we will win.
As Secretary of State, I can confidently assure you that only U.S. citizens can vote, and I take that seriously. Our reviews consistently show non-citizen voting is rare. The law is clear, safeguards are strong, and violations carry real consequences.
February 18, 2026 at 7:29 PM
As Secretary of State, I can confidently assure you that only U.S. citizens can vote, and I take that seriously. Our reviews consistently show non-citizen voting is rare. The law is clear, safeguards are strong, and violations carry real consequences.
Last week, the House voted to dismantle safeguards on weapons easily converted into fully lethal guns.
Parkland survivor and gun violence prevention advocate Aalayah Eastmond shares just how dangerous this bill is 👇
Parkland survivor and gun violence prevention advocate Aalayah Eastmond shares just how dangerous this bill is 👇
I survived the Parkland shooting. 'Less-lethal' law isn't the answer.
As Americans demand government to do something to protect us from gun violence, Congress decided to help the billion dollar weapons industry instead.
www.usatoday.com
February 17, 2026 at 8:24 PM
Last week, the House voted to dismantle safeguards on weapons easily converted into fully lethal guns.
Parkland survivor and gun violence prevention advocate Aalayah Eastmond shares just how dangerous this bill is 👇
Parkland survivor and gun violence prevention advocate Aalayah Eastmond shares just how dangerous this bill is 👇
It’s good to think about unintended consequences, sure.
But an alternate reading would be “maybe we should’ve regulated the nuclear plants AND ALSO demanded coal plants install way more emissions safeguards.
That seems preferable to a “trust me, bro” nuclear regulatory regime, y’know?
But an alternate reading would be “maybe we should’ve regulated the nuclear plants AND ALSO demanded coal plants install way more emissions safeguards.
That seems preferable to a “trust me, bro” nuclear regulatory regime, y’know?
February 15, 2026 at 7:03 PM
It’s good to think about unintended consequences, sure.
But an alternate reading would be “maybe we should’ve regulated the nuclear plants AND ALSO demanded coal plants install way more emissions safeguards.
That seems preferable to a “trust me, bro” nuclear regulatory regime, y’know?
But an alternate reading would be “maybe we should’ve regulated the nuclear plants AND ALSO demanded coal plants install way more emissions safeguards.
That seems preferable to a “trust me, bro” nuclear regulatory regime, y’know?
I always see Hasan get interrogated by journalists about his "lack of journalistic integrity and editorial safeguards"
y'all, if *this* is your journalistic integrity, i'm going to fucking dis-integrate you with the BFG from DOOM
y'all, if *this* is your journalistic integrity, i'm going to fucking dis-integrate you with the BFG from DOOM
Have you ever once seen the New York Times quote Trump like this?
February 16, 2026 at 10:09 PM
I always see Hasan get interrogated by journalists about his "lack of journalistic integrity and editorial safeguards"
y'all, if *this* is your journalistic integrity, i'm going to fucking dis-integrate you with the BFG from DOOM
y'all, if *this* is your journalistic integrity, i'm going to fucking dis-integrate you with the BFG from DOOM
“EPA’s repeal of the endangerment finding and safeguards to limit vehicle emissions marks a complete dereliction of the agency’s mission to protect people’s health and its legal obligation under the Clean Air Act,” said UCS President & CEO @gretchentg.bsky.social to @theguardian.com.
Environmental groups sue Trump’s EPA over repeal of landmark climate finding
Lawsuit from health and environmental justice groups challenges the EPA’s rollback of the ‘endangerment finding’
www.theguardian.com
February 18, 2026 at 4:41 PM
“EPA’s repeal of the endangerment finding and safeguards to limit vehicle emissions marks a complete dereliction of the agency’s mission to protect people’s health and its legal obligation under the Clean Air Act,” said UCS President & CEO @gretchentg.bsky.social to @theguardian.com.
If neoliberalism, the ideology that has dominated the Western world for ~50 years is not responsible for the failings that lead to fascism, what is?
What is to blame for all the institutions and safeguards meant to uphold American democracy crumpling like wet paper at the slightest fash touch?
What is to blame for all the institutions and safeguards meant to uphold American democracy crumpling like wet paper at the slightest fash touch?
February 16, 2026 at 2:21 PM
If neoliberalism, the ideology that has dominated the Western world for ~50 years is not responsible for the failings that lead to fascism, what is?
What is to blame for all the institutions and safeguards meant to uphold American democracy crumpling like wet paper at the slightest fash touch?
What is to blame for all the institutions and safeguards meant to uphold American democracy crumpling like wet paper at the slightest fash touch?
every couple days despite my careful safeguards some asshole tries to make me angry or otherwise care about the behavior of a streamer and i simply refuse
February 13, 2026 at 3:29 AM
every couple days despite my careful safeguards some asshole tries to make me angry or otherwise care about the behavior of a streamer and i simply refuse
What are Republicans focused on today? Creating loopholes that allow weapons be sold without background checks, serial numbers, and basic accountability safeguards.
February 12, 2026 at 5:27 PM
What are Republicans focused on today? Creating loopholes that allow weapons be sold without background checks, serial numbers, and basic accountability safeguards.
First Trump pushed out career national security experts working to counter cyber threats posed by China.
Now he’s shelving critical safeguards they put in place.
Why is the Trump Administration so eager to do Beijing’s bidding?
Now he’s shelving critical safeguards they put in place.
Why is the Trump Administration so eager to do Beijing’s bidding?
February 13, 2026 at 6:01 PM
First Trump pushed out career national security experts working to counter cyber threats posed by China.
Now he’s shelving critical safeguards they put in place.
Why is the Trump Administration so eager to do Beijing’s bidding?
Now he’s shelving critical safeguards they put in place.
Why is the Trump Administration so eager to do Beijing’s bidding?
"America often views itself as a model of democracy for the rest of the world. But other democracies have tried to put in more safeguards in place to prevent the kind of redistricting fever taking hold in the United States." (gift link)
Why Other Democracies Don’t Gerrymander Like the U.S.
www.nytimes.com
February 18, 2026 at 5:05 PM
"America often views itself as a model of democracy for the rest of the world. But other democracies have tried to put in more safeguards in place to prevent the kind of redistricting fever taking hold in the United States." (gift link)
AIの軍事利用を巡り国防総省がAnthropicとの関係解消を検討
https://gigazine.net/news/20260216-pentagon-anthropic-ai-safeguards/
https://gigazine.net/news/20260216-pentagon-anthropic-ai-safeguards/
AIの軍事利用を巡り国防総省がAnthropicとの関係解消を検討
生成AI「Claude」の軍事利用について、開発元であるAnthropicとアメリカ国防総省の交渉が難航し、国防総省側が関係解消を考えていることがわかりました。
gigazine.net
February 16, 2026 at 4:46 AM
AIの軍事利用を巡り国防総省がAnthropicとの関係解消を検討
https://gigazine.net/news/20260216-pentagon-anthropic-ai-safeguards/
https://gigazine.net/news/20260216-pentagon-anthropic-ai-safeguards/