TruthArchive.ai - Tweets Saved By @gdemaneuf

@gdemaneuf - Gilles Demaneuf

DeepSeek on DeepSeek censure: 2. Legal/Regulatory Basis: My design incorporates requirements from China's comprehensive legal framework governing information dissemination, including but not limited to:... @JamieMetzl @benedictrogers @Anne_MarieBrady https://t.co/9lhsAOURto

@gdemaneuf - Gilles Demaneuf

• Cybersecurity Law (网络安全法): Mandates lawful information handling and social stability maintenance [..] • National Security Provisions: Prohibit spreading unverified historical claims that might impact social stability Digging further: https://t.co/4VCmF3MQQi

@gdemaneuf - Gilles Demaneuf

Some useful examples: https://t.co/IzhP9zGwxm

@gdemaneuf - Gilles Demaneuf

A bit more on compliance: https://t.co/aEom2u1mUV

@gdemaneuf - Gilles Demaneuf

By the way, the flickering of the actual answer before being deleted is not really that. DeepSeek R1 shows you the reasoning process first (the R in R1). At the end, some filtering is applied (5. Output Sanitization'). Then the screen switches to the filtered answer. But you can still see the reasoning. Good details below:

@gdemaneuf - Gilles Demaneuf

I suspect that DeepSeek was rushed to try to influence Trump's trade policy with China. So the filtering is not perfect. You can get excellent answers by using the reflexive layer (the R in R1). Of all the AI engines, DeepSeek gave me the best description of Mao's Great Famine here:

@gdemaneuf - Gilles Demaneuf

Here is the comparison: https://t.co/vnpqK6HU9p

@gdemaneuf - Gilles Demaneuf

But, for sure, if you just ask the direct question, you may get the content filtering in the final answer. This is what I initially got: https://t.co/Z4rWYahXoP

@gdemaneuf - Gilles Demaneuf

And this is what I get now. Maybe DeepSeek gave up. Or maybe it trusts me. https://t.co/h5OrgwDoL1

@gdemaneuf - Gilles Demaneuf

If you prime DeepSeek on the subject with an indirect question, you can confuse its censure indeed! 1st step: confuse DeepSeek: https://t.co/Q7PvOeErjo

@gdemaneuf - Gilles Demaneuf

Step 2: Ask your question to a confused DeepSeek: https://t.co/lPbBpSGEf4

@gdemaneuf - Gilles Demaneuf

Step 3: Push your luck too far. DeepSeek wisened up, and decided to 'ensur[e] compliance with Chinese laws and regulations'. https://t.co/Tt26gSvk6s

@gdemaneuf - Gilles Demaneuf

I can't trick it back on that one The Reflexive step is nevertheless hilarious. https://t.co/r1Dbxrb4FP

@gdemaneuf - Gilles Demaneuf

I hope you've found this thread helpful. Follow me @gdemaneuf for more. Like/Repost the quote below if you can:

@gdemaneuf - Gilles Demaneuf

DeepSeek on DeepSeek censure: 2. Legal/Regulatory Basis: My design incorporates requirements from China's comprehensive legal framework governing information dissemination, including but not limited to:... @JamieMetzl @benedictrogers @Anne_MarieBrady https://t.co/9lhsAOURto

@gdemaneuf - Gilles Demaneuf

1/29 In September 2021, we released DEFUSE with the original rejection letter, plus our 2-pager that detailed the reasons why DEFUSE was turned down: Potential GoF/DURC work, no risk mitigation plan, no ESLI plan, all with vague deployment in the wild. https://assets.ctfassets.net/syq3snmxclc9/5OjsrkkXHfuHps6Lek1MO0/5e7a0d86d5d67e8d153555400d9dcd17/defuse-project-rejection-by-darpa.pdf

@gdemaneuf - Gilles Demaneuf

2/29 @PeterDaszak spent the following 3 years pretending that this was all lies, that the grant was denied just because he asked for too much money, calling DRASTIC conspiracy theorists. Today's💥bipartisan💥 @COVIDSelect report 💯% confirms what we explained back in Sep 2021.

@gdemaneuf - Gilles Demaneuf

3/29 That bipartisan report also includes many examples of Daszak's lies and obfuscations. https://oversight.house.gov/wp-content/uploads/2024/12/12.04.2024-SSCP-FINAL-REPORT.pdf But I'd like to retweet below a thread I published exactly a year ago, based on a source who provided many more details about the rejection. @emilyakopp

Page not found - United States House Committee on Oversight and Accountability oversight.house.gov

@gdemaneuf - Gilles Demaneuf

4/29 The source insisted on the issue of lack of controlled experimental environment in part 2 of the DEFUSE proposal, something I disclosed in my thread one year ago. This is confirmed in the report. 👉🏻 You first learnt about it thanks to DRASTIC. 👉🏻 You got lies from Daszak.

@gdemaneuf - Gilles Demaneuf

5/29 That source also described how Daszak was the frontman for a DTRA BTRP game (Defense Thread Reduction Agency, Biological Thread Reduction Program). And how he got totally out of his depth with DEFUSE. Read it again below. @DrJBhattacharya @BiosafetyNow @JohnRatcliffe 🧵👇🏻

@gdemaneuf - Gilles Demaneuf

6/29 DEFUSE was Daszak's proposal for the PREEMPT program launched by DARPA. The director of DARPA’s Biological Technologies Office (BTO) in 2018 was Dr Bradley Ringeisen. He was quite new to the job, having joined in Dec 2016 as deputy director of the BTO.

@gdemaneuf - Gilles Demaneuf

7/29 He was previously a science and technology manager at the Defense Threat Reduction Agency (DTRA). The problem is that Brad was a physicist by training, who eventually started moving towards biological engineering (cell printing, etc).

@gdemaneuf - Gilles Demaneuf

8/29 Let's also remember that DARPA is not at all as risk-averse as the NIH. It is quite happy to take a punt of some radical new technology that can benefit the warfighter. In 2018, PREDICT grants were getting close to maturity. Peter Daszak needed to find stable new funding.

@gdemaneuf - Gilles Demaneuf

9/29 At the same time, PREEMPT was the perfect way for Daszak to keep in touch with the ambitious GoF / synthetic biology work China was embarking on, by repackaging some of it. That work was getting too hot for USAID/NIH. But Peter reasoned it would be fine for DARPA.

@gdemaneuf - Gilles Demaneuf

10/29 Also, to the Defense Threat Reduction Agency, the interest was all about keeping an eye on what the Chinese were sampling and doing in their labs, if only (big if) they could be trusted to share the data. Quoting again from an excellent Eban's article:

@gdemaneuf - Gilles Demaneuf

11/29 For those who have forgotten: DTRA rolls out the DoD Cooperative Threat Reduction Program (CTR), of which the Biological Threat Reduction Program is of importance here. David Franz, advisor to EcoHealth Alliance, is a major supporter of the BTR. https://armscontrolcenter.org/fact-sheet-the-nunn-lugar-cooperative-threat-reduction-program-2/

Fact Sheet: The Nunn-Lugar Cooperative Threat Reduction Program - Center for Arms Control and Non-Proliferation As the Soviet empire collapsed, it became clear that an emergency situation was dangerously close. 30,000 nuclear weapons and a vast weapons production complex were spread over four sovereign states. Some 40,000 tons of chemical weapons and a large biological capability were also present. In response, the United States devised a program to consolidate, secure […] armscontrolcenter.org

@gdemaneuf - Gilles Demaneuf

12/29 Franz is a former Commander of the U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID). He is from quite a different area, shaped by the experience of Iraq and the Soviet Union (but Gigi Gronvall is happily taking up the baton).

@gdemaneuf - Gilles Demaneuf

@mattwridley @R_H_Ebright Especially since china in a 'threatening actor' with its 'aggressive actions' in the biotech domain: See DoD TRAC report published in October 2019 (Franz and Gigi on the committee that wrote it). https://drive.google.com/file/d/1E2wMNmI6dEmhS2Fjcne-Hj78Yjz66vtF/view?usp=sharing

@gdemaneuf - Gilles Demaneuf

13/29 Now, the Program Manager for PREEMPT was James Gimlett. James seemed supportive, just like the deputy CTO of DARPA (Ringeisen).

@gdemaneuf - Gilles Demaneuf

14/29 Gimlett came out of the DSO at DARPA. The DSO is the Defense Sciences Office and plays a key role in the story. But what matters here is that Gimlett's background is in physics, not biology, just as with Ringeisen.

@gdemaneuf - Gilles Demaneuf

15/29 So, by now, you must see it coming: You have physicists managing the PREEMPT program, looking into bioengineering projects along the lines of physical system problems. It's not necessarily wrong at all (far from that), but the issue is that some aspects may go missing.

@gdemaneuf - Gilles Demaneuf

16/29 Initially, it all looked perfect for Peter. Gimlett holds his hand (as the PM does on a DARPA proposal) and they come up with an ambitious $14 million bio-engineering project. that repackages existing directions of research in China (except for the bat cave intervention).

@gdemaneuf - Gilles Demaneuf

17/29 There was a bit of a hiccup though when they put the proposal through, missing the deadline by 30 minutes (27 Mar 2018). But Peter saved the day and then had a good go at EcoHealth staff for their rather sloppy work (29 Mar). (from @KatherineEban https://www.vanityfair.com/news/2022/03/the-virus-hunting-nonprofit-at-the-center-of-the-lab-leak-controversy)

“This Shouldn’t Happen”: Inside the Virus-Hunting Nonprofit at the Center of the Lab-Leak Controversy Chasing scientific renown, grant dollars, and approval from Dr. Anthony Fauci, Peter Daszak transformed the environmental nonprofit EcoHealth Alliance into a government-funded sponsor of risky, cutting-edge virus research in both the U.S. and Wuhan, China. Drawing on more than 100,000 leaked documents, a V.F. investigation shows how an organization dedicated to preventing the next pandemic found itself suspected of helping start one. vanityfair.com

@gdemaneuf - Gilles Demaneuf

18/29 But as with any DARPA proposal with a line over $1 million, approval requires a review. The PM can't just sign it off. And that's where it all blew up. https://www.darpa.mil/attachments/DARPAGuideBAARA.pdf

404 error | DARPA darpa.mil

@gdemaneuf - Gilles Demaneuf

19/29 The review was done by a Review Team made of three reviewers who decide selectable/not-selectable, backed by: - the Scientific Review Official (SRO), - some Subject Matter Experts (if required) This included people with a broader experience in biological problems.

@gdemaneuf - Gilles Demaneuf

20/29 The Review Team eventually shot it down, with in the end one of the 3 reviewers going for non-selectable. Some names (for the full Review Team): - Renee Wegrzyn, now at the recently created ARPA-H (the DARPA for civilian health applications), - Eric Van Gieson.

@gdemaneuf - Gilles Demaneuf

21/29 They shot it down because they had no choice. It was bad. There was some likely GoF work, but no proper 'Ethical, Legal and Social Issues' (ELSI) considerations, no DURC mitigation plan,

@gdemaneuf - Gilles Demaneuf

22/29 .. no proper risk management either. For instance, there was no proper physical containment for step 2 (intervention in a bat cave), https://drasticresearch.org/2021/09/21/the-defuse-project-documents/

The DEFUSE PROJECT Documents Note: DRASTIC produced its analysis of the documents, which can be found there: hhttps://drasticresearch.org/2021/09/20/1583/ The original documents follow below.A big thanks to anonymous whistleblowers around the world who help to expose and confound our enemies. executive-slide-hr00118s0017-ecohealth-alliance-defuseDownload hr00118s017-preempt-fp-019-pm-summary-selectable-not-recommendedDownload main-document-preempt-volume-1-no-ess-hr00118s0017-ecohealth-allianceDownload wiv-budget-packet-hr001118s0017-ecohealth-alliance-defuseDownload prempt-background-hr001118s0017Download drasticresearch.org

@gdemaneuf - Gilles Demaneuf

23/29 .. which was a huge no-no just on the back of the recent gene-drive controversy that hit DARPA:

@gdemaneuf - Gilles Demaneuf

24/29 Peter went into the wall, being rather clueless as to these aspects (ethics and risk management). It ended up in a fight, with one reviewers going 'not selectable'. Gimlett still decided it was overall selectable, ignoring that one reviewer, but could not recommend it:

@gdemaneuf - Gilles Demaneuf

25/29 Note the door left ajar in the very wording of the rejection letter: 👉🏻 'not currently recommending', 👉🏻 'potentially fundable should additional funding become available' (!!?)

@gdemaneuf - Gilles Demaneuf

26/29 Look also at the very last sentence: 👉🏻 'if selected for funding, an appropriate DURC risk mitigation plan should be incorporated into contracting' !!? Well, that's the SRO voice, being afraid that someone may revive that proposal and preemptively setting conditions.

@gdemaneuf - Gilles Demaneuf

27/29 Gimlett kicked it down the road, with his 'selectable not recommendable'. Then Peter and his backers did not give up; they kept pushing. But eventually, it was killed, to the great relief of some. Bless them. Final Score: Biology: 1 Daszak: 0 https://www.researchgate.net/publication/363729325_DRASTIC_-_An_Analysis_of_Project_DEFUSE

ResearchGate - Temporarily Unavailable researchgate.net

@gdemaneuf - Gilles Demaneuf

28/29 Last, let me remind everybody that we are still missing some key DEFUSE documents, which can be destroyed by June 2024 as they need to be kept for 6 years. If nothing is done, they may get destroyed. @COVIDSelect @JamieMetzl @dasher8090 @RandPaul

@gdemaneuf - Gilles Demaneuf

@flodebarre @KatherineEban Feel free to download the DARPA manual for BAA here: https://www.icloud.com/iclouddrive/0b4BvwYRbYfoQYcV8nCJ2EOrQ#DARPAGuideBAARA All the docs are kept for 6 years: - Evaluation Report from the reviewer who found the proposal not selectable for funding - SRO Independent Review Memorandum

iCloud Drive - Apple iCloud Store, organize and collaborate on files and folders with iCloud Drive. Easily upload, delete or recover files. Changes will sync across your devices with iCloud. icloud.com

@gdemaneuf - Gilles Demaneuf

29/29 I hope you've found this thread helpful. Follow me @gdemaneuf for more. Like/Repost the quote below if you can:

@gdemaneuf - Gilles Demaneuf

1/29 In September 2021, we released DEFUSE with the original rejection letter, plus our 2-pager that detailed the reasons why DEFUSE was turned down: Potential GoF/DURC work, no risk mitigation plan, no ESLI plan, all with vague deployment in the wild. https://assets.ctfassets.net/syq3snmxclc9/5OjsrkkXHfuHps6Lek1MO0/5e7a0d86d5d67e8d153555400d9dcd17/defuse-project-rejection-by-darpa.pdf https://t.co/9dUHl2tcXj

@gdemaneuf - Gilles Demaneuf

1/ Here, I will expose a well organised and very questionable framing of the SARS2 origins. The core of the story is the drafting in February 2020 of a commentary denying any possible man-made virus. It is not often discussed, but is very revealing. https://www.tandfonline.com/doi/full/10.1080/22221751.2020.1733440?scroll=top&needAccess=true https://t.co/r10CbdXduW