Based on the science, just the B. subtilis.

I still don't understand what is the real concern with B. Subtilis.

Based on the science, it has gotten a safe status and has been studied more than any probiotic bacteria in the world. I has more positive studies than negatives. It is probably the bacterial non-indigenous human species that benefits more the intestines during a dysbiosis.
It boots the immune system, help to restore the beneficial flora, and is anti-pathogenic.

What are the studies where the science has marked it as a non-safe ??

According to my research, I see the opposite. I must believe in the medical research no only when they find something negative but when they find positive results too. I can not use the medical literature to support my personal opinion but neutrally.

IMO, to say B. Subtilis isn't safe is going against the science.

How B. Subtilis got this category in Europe, Asia, and US :

. It has been marketed for over 15 years in Asia as a pharmaceutical to support gut health, B. subtilis R0179 benefits from a good track-record of safe and effective use, with over 100 post-market CLINICAL TRIALS published in adults and children. Finally this strain SAFETY has been recognised by the regulatory authorities in Europe (QSP status from EFSA), Japan, Canada and the United States where it received the self-affirmed GRAS (Generally Recognized As Safe) status in January 2012.

This is a 2008 study from UK :

J Appl Microbiol. 2008 Aug;105(2):510-20. Epub 2008 Feb 29.
The safety of Bacillus subtilis and Bacillus indicus as food probiotics.
Hong HA, Huang JM, Khaneja R, Hiep LV, Urdaci MC, Cutting SM.
Source
School of Biological Sciences, Royal Holloway, University of London, Egham, Surrey, UK.
Abstract
AIMS:
To conduct in vitro and in vivo assessments of the safety of two species of Bacillus, one of which, Bacillus subtilis, is in current use as a food supplement.
METHODS AND RESULTS:
Cultured cell lines, Caco-2, HEp-2 and the mucus-producing HT29-16E cell line, were used to evaluate adhesion, invasion and cytotoxicity. The Natto strain of B. subtilis was shown to be able to invade and lyse cells. Neither species was able to adhere significantly to any cell line. The Natto strain was also shown to form biofilms. No strain produced any of the known Bacillus enterotoxins. Disc-diffusion assays using a panel of antibiotics listed by the European Food Safety Authority (EFSA) showed that only Bacillus indicus carried resistance to clindamycin at a level above the minimum inhibitory concentration breakpoints set by the EFSA. In vivo assessments of acute and chronic dosing in guinea pigs and rabbits were made. No toxicity was observed in animals under these conditions.
CONCLUSIONS:
Bacillus indicus and B. subtilis should be considered safe for oral use although the resistance of B. indicus to clindamycin requires further study.
SIGNIFICANCE AND IMPACT OF THE STUDY:
The results support the use of B. subtilis and B. indicus strains as food supplements.

I am doing all my efforts to find information that supports B. Subtilis shouldn't be used as a Probiotic. I have read more than 20 articles supporting its benefits and safety.

This is another article from a respected microbiologist claiming B. Subtilis safety :

Bacillus subtilis – Identification & Safety
Bacillus subtilis is not an obscure or mysterious
microbe. It is, instead, a very well-studied
bacterium. It is the ‘type’ (original) species
of the Bacillus genus
(1)
and it is viewed by
microbiologists as a typical example of a
Gram-positive bacterium and an endosporeproducer. Consequently, B. subtilis attracts
a lot of research and this is why it was one
of the first organisms to have its full genome
sequenced (more than ten years ago).
So what do research studies tell us about the
safety of B. subtilis as a probiotic? Last year,
three studies
(2,3,4)
were published on the safety
of B. subtilis. Four strains were examined and
none of these strains were found to have any
pathogenic indications. The three studies
involved researchers from Canada, France,
Vietnam and the UK, including Dr Simon
Cutting, Professor of Molecular Microbiology at
Royal Holloway College, University of London.
Pathogenic genes
Specifically, the researchers tested for the
presence of genes responsible for the
production of various toxins, and other harmful
substances such as haemolysin (blood cell
disruption) and lecithinase (cell membrane
disruption). No such genes were found.
The strains were also added to gut epithelial
cells without causing harm, and were fed to
various laboratory animals (mice, rats, guinea
pigs, rabbits and piglets) with no adverse
effects.
Antibiotic resistance
Another safety consideration for probiotic
microbes is the presence of DNA that provides
protection against the effects of antibiotics.
Such antibiotic-resistant genes are found
naturally in bacteria, but the question is
whether they are resistant to antibiotics that
are used in the treatment of humans and also
whether they are the type of genes that can be
transferred to other bacteria.
If such genes are transferrable to other
bacteria, then they could be taken up
by bacteria in the human gut flora and
subsequently passed onto pathogens creating
a new type of resistant pathogen.
All three of the 2008 safety studies tested their
B. subtilis strains against a range of antibiotics.
All the strains were sensitive (not resistant)
to all the antibiotics important in medical
treatment, as listed in a report of the European
Food Safety Authority
(5)
.
Accuracy of microbe identification
Given such data confirming the safety of
B. subtilis, the major remaining question is
whether a probiotic product contains the
species it claims. Sanders et al.
(6)
reported on
three studies that examined a total of seven
Bacillus products and found that all of them
were mislabelled as to the species contained.
One product contained Bacillus cereus rather
than B. subtilis, and this is signifi cant because
some strains of B. cereus can cause food
poisoning.
The B. subtilis of Bio-Kult has been assessed
independently by The National Collection of
Industrial, Marine and Food Bacteria (NCIMB)
(www.ncimb.com) and found to be 99.7%
identical to genetic database records of
From:
Peter Cartwright BA (Hons) MA MSc
Human Microbiota Specialist
Probiotics International Ltd.
Somerset, U.K.